Description: Fix test failure with Numpy 1.24. Author: Bas Couwenberg Bug-Debian: https://bugs.debian.org/1027229 --- a/docs/source/dim_red.txt +++ b/docs/source/dim_red.txt @@ -18,10 +18,10 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 4.5], [[1,1],[1,2]], 20 # 20 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,1],[1,2]], 30 # 30 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> lda = mlpy.LDA() @@ -50,10 +50,10 @@ Example - KNN in kernel fisher space: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 4.5], [[1,1],[1,2]], 20 # 20 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,1],[1,2]], 30 # 30 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> K = mlpy.kernel_gaussian(x, x, sigma=3) # compute the kernel matrix @@ -174,7 +174,7 @@ Example: >>> np.random.seed(0) >>> np.random.seed(0) >>> x = np.zeros((150, 2)) ->>> y = np.empty(150, dtype=np.int) +>>> y = np.empty(150, dtype=int) >>> theta = np.random.normal(0, np.pi, 50) >>> r = np.random.normal(0, 0.1, 50) >>> x[0:50, 0] = r * np.cos(theta) --- a/docs/source/liblinear.txt +++ b/docs/source/liblinear.txt @@ -56,13 +56,13 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 0 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.zeros(n1, dtype=np.int) +>>> y1 = np.zeros(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class 1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = np.ones(n2, dtype=np.int) +>>> y2 = np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [5, 8], [[0.5,0],[0,0.5]], 200 # 200 samples of class 2 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 2 * np.ones(n3, dtype=np.int) +>>> y3 = 2 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> svm = mlpy.LibLinear(solver_type='l2r_l2loss_svc_dual', C=0.01) --- a/docs/source/lin_class.txt +++ b/docs/source/lin_class.txt @@ -22,10 +22,10 @@ Binary classification: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class -1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = -np.ones(n2, dtype=np.int) +>>> y2 = -np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> ldac = mlpy.LDAC() @@ -59,13 +59,13 @@ Multiclass classification: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 25], [[1,1],[1,2]], 200 # 200 samples of class 0 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.zeros(n1, dtype=np.int) +>>> y1 = np.zeros(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 22.5], [[1,0],[0,1]], 300 # 300 samples of class 1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = np.ones(n2, dtype=np.int) +>>> y2 = np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [5, 28], [[0.5,0],[0,0.5]], 200 # 200 samples of class 2 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 2 * np.ones(n3, dtype=np.int) +>>> y3 = 2 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> ldac = mlpy.LDAC() @@ -114,10 +114,10 @@ Examples >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class -1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = -np.ones(n2, dtype=np.int) +>>> y2 = -np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> p = mlpy.Perceptron(alpha=0.1, thr=0.05, maxiters=100) # basic perceptron @@ -163,10 +163,10 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class -1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = -np.ones(n2, dtype=np.int) +>>> y2 = -np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> en = mlpy.ElasticNetC(lmb=0.01, eps=0.001) @@ -219,13 +219,13 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 0 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.zeros(n1, dtype=np.int) +>>> y1 = np.zeros(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class 1 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = np.ones(n2, dtype=np.int) +>>> y2 = np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [5, 8], [[0.5,0],[0,0.5]], 200 # 200 samples of class 2 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 2 * np.ones(n3, dtype=np.int) +>>> y3 = 2 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> da = mlpy.DLDA(delta=0.1) --- a/docs/source/nonlin_class.txt +++ b/docs/source/nonlin_class.txt @@ -17,10 +17,10 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 4.5], [[1,1],[1,2]], 20 # 20 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,1],[1,2]], 30 # 30 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> K = mlpy.kernel_gaussian(x, x, sigma=2) # kernel matrix @@ -77,13 +77,13 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [5, 8], [[0.5,0],[0,0.5]], 200 # 200 samples of class 3 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 3 * np.ones(n3, dtype=np.int) +>>> y3 = 3 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> knn = mlpy.KNN(k=3) @@ -129,13 +129,13 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [6, 8], [[0.5,0],[0,0.5]], 200 # 200 samples of class 3 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 3 * np.ones(n3, dtype=np.int) +>>> y3 = 3 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> tree = mlpy.ClassTree(minsize=10) @@ -176,13 +176,13 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 5], [[1,1],[1,2]], 200 # 200 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,0],[0,1]], 300 # 300 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> mean3, cov3, n3 = [6, 8], [[0.5,0],[0,0.5]], 200 # 200 samples of class 3 >>> x3 = np.random.multivariate_normal(mean3, cov3, n3) ->>> y3 = 3 * np.ones(n3, dtype=np.int) +>>> y3 = 3 * np.ones(n3, dtype=int) >>> x = np.concatenate((x1, x2, x3), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2, y3)) >>> ml = mlpy.MaximumLikelihoodC() --- a/docs/source/svm.txt +++ b/docs/source/svm.txt @@ -123,10 +123,10 @@ Example: >>> np.random.seed(0) >>> mean1, cov1, n1 = [1, 4.5], [[1,1],[1,2]], 20 # 20 samples of class 1 >>> x1 = np.random.multivariate_normal(mean1, cov1, n1) ->>> y1 = np.ones(n1, dtype=np.int) +>>> y1 = np.ones(n1, dtype=int) >>> mean2, cov2, n2 = [2.5, 2.5], [[1,1],[1,2]], 30 # 30 samples of class 2 >>> x2 = np.random.multivariate_normal(mean2, cov2, n2) ->>> y2 = 2 * np.ones(n2, dtype=np.int) +>>> y2 = 2 * np.ones(n2, dtype=int) >>> x = np.concatenate((x1, x2), axis=0) # concatenate the samples >>> y = np.concatenate((y1, y2)) >>> K = mlpy.kernel_gaussian(x, x, sigma=2) # kernel matrix --- a/docs/source/tutorial.txt +++ b/docs/source/tutorial.txt @@ -30,7 +30,7 @@ Load the modules: Load the Iris dataset: >>> iris = np.loadtxt('iris.csv', delimiter=',') ->>> x, y = iris[:, :4], iris[:, 4].astype(np.int) # x: (observations x attributes) matrix, y: classes (1: setosa, 2: versicolor, 3: virginica) +>>> x, y = iris[:, :4], iris[:, 4].astype(int) # x: (observations x attributes) matrix, y: classes (1: setosa, 2: versicolor, 3: virginica) >>> x.shape (150, 4) >>> y.shape --- a/mlpy/adatron/adatron.pyx +++ b/mlpy/adatron/adatron.pyx @@ -72,8 +72,8 @@ class KernelAdatron: cdef np.ndarray[np.float_t, ndim=1] alpha_arr cdef double margin - K_arr = np.ascontiguousarray(K, dtype=np.float) - y_arr = np.asarray(y, dtype=np.int) + K_arr = np.ascontiguousarray(K, dtype=float) + y_arr = np.asarray(y, dtype=int) if K_arr.ndim != 2: raise ValueError("K must be a 2d array_like object") @@ -93,7 +93,7 @@ class KernelAdatron: ynew = np.where(y_arr==self._labels[0], -1, 1) n = K_arr.shape[0] - alpha_arr = np.zeros(n, dtype=np.float) + alpha_arr = np.zeros(n, dtype=float) steps = adatron( ynew.data, K_arr.data, n, self._C, self._maxsteps, @@ -121,7 +121,7 @@ class KernelAdatron: if self._alpha is None: raise ValueError("no model computed; run learn() first") - Kt_arr = np.asarray(Kt, dtype=np.float) + Kt_arr = np.asarray(Kt, dtype=float) try: s = np.sign(np.dot(self._alpha * self._y, Kt_arr.T)) @@ -129,7 +129,7 @@ class KernelAdatron: raise ValueError("Kt, alpha: shape mismatch") return np.where(s==-1, self._labels[0], self._labels[1]) \ - .astype(np.int) + .astype(int) def margin(self): """Return the margin. --- a/mlpy/bordacount/borda.py +++ b/mlpy/bordacount/borda.py @@ -59,7 +59,7 @@ def borda_count(x, k=None): * ... """ - x_arr = np.asarray(x, dtype=np.int) + x_arr = np.asarray(x, dtype=int) n, p = x_arr.shape if k == None: --- a/mlpy/canberra/canberra.pyx +++ b/mlpy/canberra/canberra.pyx @@ -16,8 +16,8 @@ def canberra(x, y): cdef np.ndarray[np.float64_t, ndim=1] x_arr cdef np.ndarray[np.float64_t, ndim=1] y_arr - x_arr = np.ascontiguousarray(x, dtype=np.float) - y_arr = np.ascontiguousarray(y, dtype=np.float) + x_arr = np.ascontiguousarray(x, dtype=float) + y_arr = np.ascontiguousarray(y, dtype=float) if x_arr.shape[0] != y_arr.shape[0]: raise ValueError("x, y: shape mismatch") @@ -36,8 +36,8 @@ def canberra_location(x, y, k=None): cdef np.ndarray[np.int64_t, ndim=1] x_arr cdef np.ndarray[np.int64_t, ndim=1] y_arr - x_arr = np.ascontiguousarray(x, dtype=np.int) - y_arr = np.ascontiguousarray(y, dtype=np.int) + x_arr = np.ascontiguousarray(x, dtype=int) + y_arr = np.ascontiguousarray(y, dtype=int) if x_arr.shape[0] != y_arr.shape[0]: raise ValueError("x, y: shape mismatch") @@ -75,7 +75,7 @@ def canberra_stability(x, k=None): cdef np.ndarray[np.int64_t, ndim=2] x_arr - x_arr = np.ascontiguousarray(x, dtype=np.int) + x_arr = np.ascontiguousarray(x, dtype=int) if k == None: k = x_arr.shape[1] --- a/mlpy/crossval.py +++ b/mlpy/crossval.py @@ -73,11 +73,11 @@ def cv_kfold(n, k, strat=None, seed=0): raise ValueError("k must be > 1") if strat is not None: - _strat = np.asarray(strat, dtype=np.int) + _strat = np.asarray(strat, dtype=int) if n != _strat.shape[0]: raise ValueError("a, strat: shape mismatch") else: - _strat = np.zeros(n, dtype=np.int) + _strat = np.zeros(n, dtype=int) labels = np.unique(_strat) @@ -152,11 +152,11 @@ def cv_random(n, k, p, strat=None, seed= raise ValueError("p must be in [0, 100]") if strat is not None: - _strat = np.asarray(strat, dtype=np.int) + _strat = np.asarray(strat, dtype=int) if n != _strat.shape[0]: raise ValueError("a, strat: shape mismatch") else: - _strat = np.zeros(n, dtype=np.int) + _strat = np.zeros(n, dtype=int) labels = np.unique(_strat) --- a/mlpy/da.py +++ b/mlpy/da.py @@ -43,8 +43,8 @@ class LDAC: target values (N) """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -58,9 +58,9 @@ class LDAC: if k < 2: raise ValueError("number of classes must be >= 2") - p = np.empty(k, dtype=np.float) - mu = np.empty((k, xarr.shape[1]), dtype=np.float) - cov = np.zeros((xarr.shape[1], xarr.shape[1]), dtype=np.float) + p = np.empty(k, dtype=float) + mu = np.empty((k, xarr.shape[1]), dtype=float) + cov = np.zeros((xarr.shape[1], xarr.shape[1]), dtype=float) for i in range(k): wi = (yarr == self._labels[i]) @@ -71,8 +71,8 @@ class LDAC: cov /= float(xarr.shape[0] - k) covinv = np.linalg.inv(cov) - self._w = np.empty((k, xarr.shape[1]), dtype=np.float) - self._bias = np.empty(k, dtype=np.float) + self._w = np.empty((k, xarr.shape[1]), dtype=float) + self._bias = np.empty(k, dtype=float) for i in range(k): self._w[i] = np.dot(covinv, mu[i]) @@ -131,16 +131,16 @@ class LDAC: if self._w is None: raise ValueError("no model computed.") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim == 1: - delta = np.empty(self._labels.shape[0], dtype=np.float) + delta = np.empty(self._labels.shape[0], dtype=float) for i in range(self._labels.shape[0]): delta[i] = np.dot(tarr, self._w[i]) + self._bias[i] return self._labels[np.argmax(delta)] else: delta = np.empty((tarr.shape[0], self._labels.shape[0]), - dtype=np.float) + dtype=float) for i in range(self._labels.shape[0]): delta[:, i] = np.dot(tarr, self._w[i]) + self._bias[i] return self._labels[np.argmax(delta, axis=1)] @@ -178,8 +178,8 @@ class DLDA: target values (N) """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -196,10 +196,10 @@ class DLDA: xm = np.mean(xarr, axis=0) self._xstd = np.std(xarr, axis=0, ddof=1) s0 = np.median(self._xstd) - self._dprime = np.empty((k, xarr.shape[1]), dtype=np.float) - self._xmprime = np.empty((k, xarr.shape[1]), dtype=np.float) + self._dprime = np.empty((k, xarr.shape[1]), dtype=float) + self._xmprime = np.empty((k, xarr.shape[1]), dtype=float) n = yarr.shape[0] - self._p = np.empty(k, dtype=np.float) + self._p = np.empty(k, dtype=float) for i in range(k): yi = (yarr == self._labels[i]) @@ -264,12 +264,12 @@ class DLDA: if self._xmprime is None: raise ValueError("no model computed.") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim == 1: return self._labels[np.argmax(self._score(tarr))] else: - ret = np.empty(tarr.shape[0], dtype=np.int) + ret = np.empty(tarr.shape[0], dtype=int) for i in range(tarr.shape[0]): ret[i] = self._labels[np.argmax(self._score(tarr[i]))] return ret @@ -282,13 +282,13 @@ class DLDA: if self._xmprime is None: raise ValueError("no model computed.") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim == 1: return self._prob(tarr) else: ret = np.empty((tarr.shape[0], self._labels.shape[0]), - dtype=np.float) + dtype=float) for i in range(tarr.shape[0]): ret[i] = self._prob(tarr[i]) return ret @@ -337,8 +337,8 @@ class KFDAC: class labels (only two classes) """ - Karr = np.asarray(K, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + Karr = np.asarray(K, dtype=float) + yarr = np.asarray(y, dtype=int) if Karr.ndim != 2: raise ValueError("K must be a 2d array_like object") @@ -405,7 +405,7 @@ class KFDAC: if self._alpha is None: raise ValueError("no model computed; run learn()") - Ktarr = np.asarray(Kt, dtype=np.float) + Ktarr = np.asarray(Kt, dtype=float) if self._kernel is not None: Ktarr = self._kernel.kernel(Ktarr, self._x) @@ -415,7 +415,7 @@ class KFDAC: raise ValueError("Kt, alpha: shape mismatch") return np.where(s==1, self._labels[0], self._labels[1]) \ - .astype(np.int) + .astype(int) def alpha(self): """Return alpha. --- a/mlpy/dimred.py +++ b/mlpy/dimred.py @@ -76,7 +76,7 @@ def lda(xarr, yarr): n, p = xarr.shape[0], xarr.shape[1] labels = np.unique(yarr) - sw = np.zeros((p, p), dtype=np.float) + sw = np.zeros((p, p), dtype=float) for i in labels: idx = np.where(yarr==i)[0] sw += np.cov(xarr[idx], rowvar=0) * \ @@ -121,7 +121,7 @@ def srda(xarr, yarr, alpha): # Point 1 in section 4.2 yu = np.unique(yarr) - yk = np.zeros((yu.shape[0]+1, yarr.shape[0]), dtype=np.float) + yk = np.zeros((yu.shape[0]+1, yarr.shape[0]), dtype=float) yk[0] = 1. for i in range(1, yk.shape[0]): yk[i][yarr==yu[i-1]] = 1. @@ -129,7 +129,7 @@ def srda(xarr, yarr, alpha): yk = yk[1:-1] # Point 2 in section 4.2 - ak = np.empty((yk.shape[0], xarr.shape[1]), dtype=np.float) + ak = np.empty((yk.shape[0], xarr.shape[1]), dtype=float) for i in range(yk.shape[0]): ak[i] = ridge_base(xarr, yk[i], alpha) @@ -258,14 +258,14 @@ def lda_fast(xarr, yarr): """ yu = np.unique(yarr) - yk = np.zeros((yu.shape[0]+1, yarr.shape[0]), dtype=np.float) + yk = np.zeros((yu.shape[0]+1, yarr.shape[0]), dtype=float) yk[0] = 1. for i in range(1, yk.shape[0]): yk[i][yarr==yu[i-1]] = 1. gso(yk, norm=False) # orthogonalize yk yk = yk[1:-1] - ak = np.empty((yk.shape[0], xarr.shape[1]), dtype=np.float) + ak = np.empty((yk.shape[0], xarr.shape[1]), dtype=float) for i in range(yk.shape[0]): ak[i], _ = ols_base(xarr, yk[i], -1) @@ -379,8 +379,8 @@ class LDA: variable, while the rows contain observations. """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -405,7 +405,7 @@ class LDA: if self._coeff is None: raise ValueError("no model computed") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) try: return np.dot(tarr-self._mean, self._coeff) @@ -444,8 +444,8 @@ class SRDA: variable, while the rows contain observations. """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -467,7 +467,7 @@ class SRDA: if self._coeff is None: raise ValueError("no model computed") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) try: return np.dot(tarr-self._mean, self._coeff) @@ -521,8 +521,8 @@ class KFDA: class labels (only two classes) """ - Karr = np.array(K, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + Karr = np.array(K, dtype=float) + yarr = np.asarray(y, dtype=int) if yarr.ndim != 1: raise ValueError("y must be an 1d array_like object") @@ -551,7 +551,7 @@ class KFDA: if self._coeff is None: raise ValueError("no model computed") - Ktarr = np.asarray(Kt, dtype=np.float) + Ktarr = np.asarray(Kt, dtype=float) if self._kernel is not None: Ktarr = self._kernel.kernel(Ktarr, self._x) @@ -595,7 +595,7 @@ class PCA: variable, while the rows contain observations. """ - xarr = np.asarray(x, dtype=np.float) + xarr = np.asarray(x, dtype=float) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -604,7 +604,7 @@ class PCA: if self._whiten: self._coeff_inv = np.empty((self._coeff.shape[1], - self._coeff.shape[0]), dtype=np.float) + self._coeff.shape[0]), dtype=float) for i in range(len(self._evals)): eval_sqrt = np.sqrt(self._evals[i]) @@ -630,7 +630,7 @@ class PCA: raise ValueError("k must be in [1, %d] or None" % \ self._coeff.shape[1]) - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) try: return np.dot(tarr-self._mean, self._coeff[:, :k]) @@ -645,7 +645,7 @@ class PCA: if self._coeff is None: raise ValueError("no PCA computed") - zarr = np.asarray(z, dtype=np.float) + zarr = np.asarray(z, dtype=float) return np.dot(zarr, self._coeff_inv[:zarr.shape[1]]) +\ self._mean @@ -699,7 +699,7 @@ class PCAFast: variable, while the rows contain observations. """ - xarr = np.asarray(x, dtype=np.float) + xarr = np.asarray(x, dtype=float) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -715,7 +715,7 @@ class PCAFast: if self._coeff is None: raise ValueError("no PCA computed") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) try: return np.dot(tarr-self._mean, self._coeff) @@ -730,7 +730,7 @@ class PCAFast: if self._coeff is None: raise ValueError("no PCA computed") - zarr = np.asarray(z, dtype=np.float) + zarr = np.asarray(z, dtype=float) return np.dot(zarr, self._coeff_inv) + self._mean def coeff(self): @@ -784,7 +784,7 @@ class KPCA: training data in input space (if kernel is a Kernel object) """ - Karr = np.asarray(K, dtype=np.float) + Karr = np.asarray(K, dtype=float) if Karr.ndim != 2: raise ValueError("K must be a 2d array_like object") @@ -818,7 +818,7 @@ class KPCA: raise ValueError("k must be in [1, %d] or None" % \ self._coeff.shape[1]) - Ktarr = np.asarray(Kt, dtype=np.float) + Ktarr = np.asarray(Kt, dtype=float) if self._kernel is not None: Ktarr = self._kernel.kernel(Ktarr, self._x) --- a/mlpy/dtw/dtw.pyx +++ b/mlpy/dtw/dtw.pyx @@ -62,9 +62,9 @@ def dtw_std(x, y, dist_only=True, square cdef int i cdef int sq - x_arr = np.ascontiguousarray(x, dtype=np.float) - y_arr = np.ascontiguousarray(y, dtype=np.float) - cost_arr = np.empty((x_arr.shape[0], y_arr.shape[0]), dtype=np.float) + x_arr = np.ascontiguousarray(x, dtype=float) + y_arr = np.ascontiguousarray(y, dtype=float) + cost_arr = np.empty((x_arr.shape[0], y_arr.shape[0]), dtype=float) if squared: sq = 1 else: sq = 0 @@ -77,8 +77,8 @@ def dtw_std(x, y, dist_only=True, square else: path( cost_arr.data, cost_arr.shape[0], cost_arr.shape[1], -1, -1, &p) - px_arr = np.empty(p.k, dtype=np.int) - py_arr = np.empty(p.k, dtype=np.int) + px_arr = np.empty(p.k, dtype=int) + py_arr = np.empty(p.k, dtype=int) for i in range(p.k): px_arr[i] = p.px[i] py_arr[i] = p.py[i] @@ -122,9 +122,9 @@ def dtw_subsequence(x, y): cdef Path p cdef int i - x_arr = np.ascontiguousarray(x, dtype=np.float) - y_arr = np.ascontiguousarray(y, dtype=np.float) - cost_arr = np.empty((x_arr.shape[0], y_arr.shape[0]), dtype=np.float) + x_arr = np.ascontiguousarray(x, dtype=float) + y_arr = np.ascontiguousarray(y, dtype=float) + cost_arr = np.empty((x_arr.shape[0], y_arr.shape[0]), dtype=float) subsequence( x_arr.data, y_arr.data, x_arr.shape[0], y_arr.shape[0], @@ -136,8 +136,8 @@ def dtw_subsequence(x, y): subsequence_path( cost_arr.data, x_arr.shape[0], y_arr.shape[0], idx, &p) - px_arr = np.empty(p.k, dtype=np.int) - py_arr = np.empty(p.k, dtype=np.int) + px_arr = np.empty(p.k, dtype=int) + py_arr = np.empty(p.k, dtype=int) for i in range(p.k): px_arr[i] = p.px[i] --- a/mlpy/elasticnet.py +++ b/mlpy/elasticnet.py @@ -86,7 +86,7 @@ def elasticnet_base(x, y, lmb, eps, supp step = 1. / (np.linalg.eigvalsh(xx).max() * 1.1) lmb = lmb * n * step damp = 1. / (1 + lmb * eps) - beta0 = np.zeros(p, dtype=np.float) + beta0 = np.zeros(p, dtype=float) k, i = 0, 0 kmax = 100000 @@ -166,8 +166,8 @@ class ElasticNet(object): response """ - xarr = np.array(x, dtype=np.float, copy=True) - yarr = np.array(y, dtype=np.float, copy=True) + xarr = np.array(x, dtype=float, copy=True) + yarr = np.array(y, dtype=float, copy=True) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -209,7 +209,7 @@ class ElasticNet(object): if self._beta0 is None: raise ValueError('no mode computed; run learn() first') - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2 or tarr.ndim < 1: raise ValueError("t must be an 1d or a 2d array_like object") @@ -283,7 +283,7 @@ class ElasticNetC(ElasticNet): class labels """ - yarr = np.asarray(y, dtype=np.int) + yarr = np.asarray(y, dtype=int) self._labels = np.unique(yarr) k = self._labels.shape[0] --- a/mlpy/findpeaks/findpeaks.pyx +++ b/mlpy/findpeaks/findpeaks.pyx @@ -61,9 +61,9 @@ def findpeaks_win(x, span): if (span % 2 == 0) or (span < 3): raise ValueError("span must be >= 3 and odd") - xarr = np.ascontiguousarray(x, dtype=np.float) + xarr = np.ascontiguousarray(x, dtype=float) retc = fp_win( xarr.data, xarr.shape[0], span, &m) - ret = np.empty(m, dtype=np.int) + ret = np.empty(m, dtype=int) for i in range(m): ret[i] = retc[i] @@ -107,7 +107,7 @@ def findpeaks_dist(x, mindist=2): if mindist < 2: raise ValueError("mindist must be >= 2") - _x = np.ascontiguousarray(x, dtype=np.float) + _x = np.ascontiguousarray(x, dtype=float) idx = findpeaks_win(_x, 3) tmp = np.empty_like(idx) --- a/mlpy/golub.py +++ b/mlpy/golub.py @@ -47,8 +47,8 @@ class Golub: target values (N) """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -85,7 +85,7 @@ class Golub: if self._w is None: raise ValueError("no model computed") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2: raise ValueError("t must be an 1d or a 2d array_like object") @@ -95,7 +95,7 @@ class Golub: raise ValueError("t, model: shape mismatch") return np.where(tmp>0, self._labels[1], self._labels[0]) \ - .astype(np.int) + .astype(int) def w(self): """Returns the coefficients. --- a/mlpy/gsl/gsl.pyx +++ b/mlpy/gsl/gsl.pyx @@ -12,7 +12,7 @@ def stats_quantile_from_sorted_data (sor cdef np.ndarray[np.float_t, ndim=1] sorted_data_arr try: - sorted_data_arr = np.array(sorted_data, dtype=np.float, order='C') + sorted_data_arr = np.array(sorted_data, dtype=float, order='C') except ValueError: raise ValueError("sorted_data must be a 2d array_like object") --- a/mlpy/kernel/kernel.pyx +++ b/mlpy/kernel/kernel.pyx @@ -33,15 +33,15 @@ def kernel_linear(t, x): cdef np.ndarray[np.float64_t, ndim=2] k_arr cdef int i, j, nt, nx, pt, px - t_arr = np.array(t, dtype=np.float, order='C', ndmin=2) - x_arr = np.array(x, dtype=np.float, order='C', ndmin=2) + t_arr = np.array(t, dtype=float, order='C', ndmin=2) + x_arr = np.array(x, dtype=float, order='C', ndmin=2) nt, pt = t_arr.shape[0], t_arr.shape[1] nx, px = x_arr.shape[0], x_arr.shape[1] if pt != px: raise ValueError("t, x: shape mismatch") - k_arr = np.empty((nt, nx), dtype=np.float) + k_arr = np.empty((nt, nx), dtype=float) for i in range(nt): for j in range(nx): k_arr[i, j] = linear( t_arr.data + (i * pt), @@ -60,15 +60,15 @@ def kernel_polynomial(t, x, gamma=1.0, b cdef np.ndarray[np.float64_t, ndim=2] k_arr cdef int i, j, nt, nx, pt, px - t_arr = np.array(t, dtype=np.float, order='C', ndmin=2) - x_arr = np.array(x, dtype=np.float, order='C', ndmin=2) + t_arr = np.array(t, dtype=float, order='C', ndmin=2) + x_arr = np.array(x, dtype=float, order='C', ndmin=2) nt, pt = t_arr.shape[0], t_arr.shape[1] nx, px = x_arr.shape[0], x_arr.shape[1] if pt != px: raise ValueError("t, x: shape mismatch") - k_arr = np.empty((nt, nx), dtype=np.float) + k_arr = np.empty((nt, nx), dtype=float) for i in range(nt): for j in range(nx): k_arr[i, j] = polynomial( t_arr.data + (i * pt), @@ -87,15 +87,15 @@ def kernel_gaussian(t, x, sigma=1.0): cdef np.ndarray[np.float64_t, ndim=2] k_arr cdef int i, j, nt, nx, pt, px - t_arr = np.array(t, dtype=np.float, order='C', ndmin=2) - x_arr = np.array(x, dtype=np.float, order='C', ndmin=2) + t_arr = np.array(t, dtype=float, order='C', ndmin=2) + x_arr = np.array(x, dtype=float, order='C', ndmin=2) nt, pt = t_arr.shape[0], t_arr.shape[1] nx, px = x_arr.shape[0], x_arr.shape[1] if pt != px: raise ValueError("t, x: shape mismatch") - k_arr = np.empty((nt, nx), dtype=np.float) + k_arr = np.empty((nt, nx), dtype=float) for i in range(nt): for j in range(nx): k_arr[i, j] = gaussian( t_arr.data + (i * pt), @@ -114,15 +114,15 @@ def kernel_exponential(t, x, sigma=1.0): cdef np.ndarray[np.float64_t, ndim=2] k_arr cdef int i, j, nt, nx, pt, px - t_arr = np.array(t, dtype=np.float, order='C', ndmin=2) - x_arr = np.array(x, dtxpe=np.float, order='C', ndmin=2) + t_arr = np.array(t, dtype=float, order='C', ndmin=2) + x_arr = np.array(x, dtxpe=float, order='C', ndmin=2) nt, pt = t_arr.shape[0], t_arr.shape[1] nx, px = x_arr.shape[0], x_arr.shape[1] if pt != px: raise ValueError("t, x: shape mismatch") - k_arr = np.empty((nt, nx), dtype=np.float) + k_arr = np.empty((nt, nx), dtype=float) for i in range(nt): for j in range(nx): k_arr[i, j] = exponential( t_arr.data + (i * pt), @@ -141,15 +141,15 @@ def kernel_sigmoid(t, x, gamma=1.0, b=1. cdef np.ndarray[np.float64_t, ndim=2] k_arr cdef int i, j, nt, nx, pt, px - t_arr = np.array(t, dtype=np.float, order='C', ndmin=2) - x_arr = np.array(x, dtype=np.float, order='C', ndmin=2) + t_arr = np.array(t, dtype=float, order='C', ndmin=2) + x_arr = np.array(x, dtype=float, order='C', ndmin=2) nt, pt = t_arr.shape[0], t_arr.shape[1] nx, px = x_arr.shape[0], x_arr.shape[1] if pt != px: raise ValueError("t, x: shape mismatch") - k_arr = np.empty((nt, nx), dtype=np.float) + k_arr = np.empty((nt, nx), dtype=float) for i in range(nt): for j in range(nx): k_arr[i, j] = sigmoid( t_arr.data + (i * pt), @@ -165,8 +165,8 @@ def kernel_center(Kt, K): """Kernel matrix centering. """ - Kt_arr = np.array(Kt, dtype=np.float) - K_arr = np.array(K, dtype=np.float) + Kt_arr = np.array(Kt, dtype=float) + K_arr = np.array(K, dtype=float) if Kt_arr.ndim == 1: J1 = np.mean(Kt_arr) --- a/mlpy/kmeans/kmeans.pyx +++ b/mlpy/kmeans/kmeans.pyx @@ -31,12 +31,12 @@ def kmeans(x, k, plus=False, seed=0): cdef np.ndarray[np.float64_t, ndim=2] means_arr cdef np.ndarray[np.int32_t, ndim=1] cls_arr - x_arr = np.ascontiguousarray(x, dtype=np.float) + x_arr = np.ascontiguousarray(x, dtype=float) if k <= 1 or k >= x_arr.shape[0]: raise ValueError("k must be in [2, N-1]") - means_arr = np.empty((k, x_arr.shape[1]), dtype=np.float) + means_arr = np.empty((k, x_arr.shape[1]), dtype=float) cls_arr = np.empty(x_arr.shape[0], dtype=np.int32) if plus: @@ -52,4 +52,4 @@ def kmeans(x, k, plus=False, seed=0): cls_arr.data, x_arr.shape[0], x_arr.shape[1], k) - return cls_arr.astype(np.int), means_arr, steps + return cls_arr.astype(int), means_arr, steps --- a/mlpy/lars.py +++ b/mlpy/lars.py @@ -52,8 +52,8 @@ def lars_base(x, y, maxsteps=None): mu = np.ones(xarr.shape[0]) active = [] inactive = range(xarr.shape[1]) - beta = np.zeros(xarr.shape[1], dtype=np.float) - est = np.zeros((maxsteps+1, xarr.shape[1]), dtype=np.float) + beta = np.zeros(xarr.shape[1], dtype=float) + est = np.zeros((maxsteps+1, xarr.shape[1]), dtype=float) for i in range(maxsteps): @@ -140,8 +140,8 @@ class LARS(): response """ - xarr = np.array(x, dtype=np.float, copy=True) - yarr = np.array(y, dtype=np.float, copy=True) + xarr = np.array(x, dtype=float, copy=True) + yarr = np.array(y, dtype=float, copy=True) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -183,7 +183,7 @@ class LARS(): if not self._beta or not self._beta0: raise ValueError('no mode computed; run learn() first') - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2 or tarr.ndim < 1: raise ValueError("t must be an 1d or a 2d array_like object") --- a/mlpy/lcs/lcs.pyx +++ b/mlpy/lcs/lcs.pyx @@ -68,8 +68,8 @@ def lcs_std(x, y): cdef Path p cdef int length - x_arr = np.ascontiguousarray(x, dtype=np.int) - y_arr = np.ascontiguousarray(y, dtype=np.int) + x_arr = np.ascontiguousarray(x, dtype=int) + y_arr = np.ascontiguousarray(y, dtype=int) b = malloc ((x_arr.shape[0]+1) * sizeof(char *)) for i in range(x_arr.shape[0]+1): @@ -84,8 +84,8 @@ def lcs_std(x, y): free (b[i]) free(b) - px_arr = np.empty(p.k, dtype=np.int) - py_arr = np.empty(p.k, dtype=np.int) + px_arr = np.empty(p.k, dtype=int) + py_arr = np.empty(p.k, dtype=int) for i in range(p.k): px_arr[i] = p.px[i] @@ -132,8 +132,8 @@ def lcs_real(x, y, eps, delta): cdef Path p cdef int length - x_arr = np.ascontiguousarray(x, dtype=np.float) - y_arr = np.ascontiguousarray(y, dtype=np.float) + x_arr = np.ascontiguousarray(x, dtype=float) + y_arr = np.ascontiguousarray(y, dtype=float) b = malloc ((x_arr.shape[0]+1) * sizeof(char *)) for i in range(x_arr.shape[0]+1): @@ -149,8 +149,8 @@ def lcs_real(x, y, eps, delta): free (b[i]) free(b) - px_arr = np.empty(p.k, dtype=np.int) - py_arr = np.empty(p.k, dtype=np.int) + px_arr = np.empty(p.k, dtype=int) + py_arr = np.empty(p.k, dtype=int) for i in range(p.k): px_arr[i] = p.px[i] --- a/mlpy/liblinear/liblinear.pyx +++ b/mlpy/liblinear/liblinear.pyx @@ -221,7 +221,7 @@ cdef class LibLinear: p = predict(self.model, test_node) free(test_node) else: - p = np.empty(tarr.shape[0], dtype=np.int) + p = np.empty(tarr.shape[0], dtype=int) for i in range(tarr.shape[0]): test_node = array1d_to_node(tarr[i]) p[i] = predict(self.model, test_node) @@ -267,14 +267,14 @@ cdef class LibLinear: dec_values = malloc (n * sizeof(double)) if tarr.ndim == 1: - dec_values_arr = np.empty(n, dtype=np.float) + dec_values_arr = np.empty(n, dtype=float) test_node = array1d_to_node(tarr) p = predict_values(self.model, test_node, dec_values) free(test_node) for j in range(n): dec_values_arr[j] = dec_values[j] else: - dec_values_arr = np.empty((tarr.shape[0], n), dtype=np.float) + dec_values_arr = np.empty((tarr.shape[0], n), dtype=float) for i in range(tarr.shape[0]): test_node = array1d_to_node(tarr[i]) p = predict_values(self.model, test_node, dec_values) @@ -316,7 +316,7 @@ cdef class LibLinear: sizeof(double)) if tarr.ndim == 1: - prob_estimates_arr = np.empty(self.model.nr_class, dtype=np.float) + prob_estimates_arr = np.empty(self.model.nr_class, dtype=float) test_node = array1d_to_node(tarr) p = predict_probability(self.model, test_node, prob_estimates) free(test_node) @@ -324,7 +324,7 @@ cdef class LibLinear: prob_estimates_arr[j] = prob_estimates[j] else: prob_estimates_arr = np.empty((tarr.shape[0], self.model.nr_class), - dtype=np.float) + dtype=float) for i in range(tarr.shape[0]): test_node = array1d_to_node(tarr[i]) p = predict_probability(self.model, test_node, prob_estimates) --- a/mlpy/libml/libml.pyx +++ b/mlpy/libml/libml.pyx @@ -63,8 +63,8 @@ cdef class KNN: cdef double **xpp cdef int i - xarr = np.ascontiguousarray(x, dtype=np.float) - yarr = np.ascontiguousarray(y, dtype=np.int) + xarr = np.ascontiguousarray(x, dtype=float) + yarr = np.ascontiguousarray(y, dtype=int) if self.k > xarr.shape[0]: raise ValueError("k must be smaller than number of samples") @@ -126,7 +126,7 @@ cdef class KNN: if self.nn.x is NULL: raise ValueError("no model computed") - tarr = np.ascontiguousarray(t, dtype=np.float) + tarr = np.ascontiguousarray(t, dtype=float) if tarr.ndim > 2: raise ValueError("t must be an 1d or a 2d array_like object") @@ -149,7 +149,7 @@ cdef class KNN: ret = self.classes[p-1] else: - ret = np.empty(tarr.shape[0], dtype=np.int) + ret = np.empty(tarr.shape[0], dtype=int) for i in range(tarr.shape[0]): tiarr = tarr[i] tdata = tiarr.data @@ -184,7 +184,7 @@ cdef class KNN: if self.nn.x is NULL: raise ValueError("no model computed") - ret = np.empty(self.nn.nclasses, dtype=np.int) + ret = np.empty(self.nn.nclasses, dtype=int) for i in range(self.nn.nclasses): ret[i] = self.classes[i] @@ -257,8 +257,8 @@ cdef class ClassTree: cdef double **xpp cdef int i - xarr = np.ascontiguousarray(x, dtype=np.float) - yarr = np.ascontiguousarray(y, dtype=np.int) + xarr = np.ascontiguousarray(x, dtype=float) + yarr = np.ascontiguousarray(y, dtype=int) yu = np.unique(yarr) if yu.shape[0] <= 1: @@ -317,7 +317,7 @@ cdef class ClassTree: if self.tree.x is NULL: raise ValueError("no model computed") - tarr = np.ascontiguousarray(t, dtype=np.float) + tarr = np.ascontiguousarray(t, dtype=float) if tarr.ndim > 2: raise ValueError("t must be an 1d or a 2d array_like object") @@ -340,7 +340,7 @@ cdef class ClassTree: ret = self.classes[p-1] else: - ret = np.empty(tarr.shape[0], dtype=np.int) + ret = np.empty(tarr.shape[0], dtype=int) for i in range(tarr.shape[0]): tiarr = tarr[i] tdata = tiarr.data @@ -375,7 +375,7 @@ cdef class ClassTree: if self.tree.x is NULL: raise ValueError("no model computed") - ret = np.empty(self.tree.nclasses, dtype=np.int) + ret = np.empty(self.tree.nclasses, dtype=int) for i in range(self.tree.nclasses): ret[i] = self.classes[i] @@ -452,8 +452,8 @@ cdef class MaximumLikelihoodC: cdef double **xpp cdef int i - xarr = np.ascontiguousarray(x, dtype=np.float) - yarr = np.ascontiguousarray(y, dtype=np.int) + xarr = np.ascontiguousarray(x, dtype=float) + yarr = np.ascontiguousarray(y, dtype=int) yu = np.unique(yarr) if yu.shape[0] <= 1: @@ -512,7 +512,7 @@ cdef class MaximumLikelihoodC: if self.ml.mean is NULL: raise ValueError("no model computed") - tarr = np.ascontiguousarray(t, dtype=np.float) + tarr = np.ascontiguousarray(t, dtype=float) if tarr.ndim > 2: raise ValueError("t must be an 1d or a 2d array_like object") @@ -535,7 +535,7 @@ cdef class MaximumLikelihoodC: ret = self.classes[p-1] else: - ret = np.empty(tarr.shape[0], dtype=np.int) + ret = np.empty(tarr.shape[0], dtype=int) for i in range(tarr.shape[0]): tiarr = tarr[i] tdata = tiarr.data @@ -570,7 +570,7 @@ cdef class MaximumLikelihoodC: if self.ml.mean is NULL: raise ValueError("no model computed") - ret = np.empty(self.ml.nclasses, dtype=np.int) + ret = np.empty(self.ml.nclasses, dtype=int) for i in range(self.ml.nclasses): ret[i] = self.classes[i] --- a/mlpy/libsvm/libsvm.pyx +++ b/mlpy/libsvm/libsvm.pyx @@ -315,14 +315,14 @@ cdef class LibSvm: dec_values = malloc (n * sizeof(double)) if tarr.ndim == 1: - dec_values_arr = np.empty(n, dtype=np.float) + dec_values_arr = np.empty(n, dtype=float) test_node = array1d_to_node(tarr) p = svm_predict_values(self.model, test_node, dec_values) free(test_node) for j in range(n): dec_values_arr[j] = dec_values[j] else: - dec_values_arr = np.empty((tarr.shape[0], n), dtype=np.float) + dec_values_arr = np.empty((tarr.shape[0], n), dtype=float) for i in range(tarr.shape[0]): test_node = array1d_to_node(tarr[i]) p = svm_predict_values(self.model, test_node, dec_values) @@ -375,7 +375,7 @@ cdef class LibSvm: sizeof(double)) if tarr.ndim == 1: - prob_estimates_arr = np.empty(self.model.nr_class, dtype=np.float) + prob_estimates_arr = np.empty(self.model.nr_class, dtype=float) test_node = array1d_to_node(tarr) p = svm_predict_probability(self.model, test_node, prob_estimates) @@ -384,7 +384,7 @@ cdef class LibSvm: prob_estimates_arr[j] = prob_estimates[j] else: prob_estimates_arr = np.empty((tarr.shape[0], self.model.nr_class), - dtype=np.float) + dtype=float) for i in range(tarr.shape[0]): test_node = array1d_to_node(tarr[i]) p = svm_predict_probability(self.model, test_node, --- a/mlpy/metrics.py +++ b/mlpy/metrics.py @@ -74,8 +74,8 @@ def error(t, p): error : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -97,8 +97,8 @@ def accuracy(t, p): accuracy : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -124,8 +124,8 @@ def error_p(t, p): errorp : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -161,8 +161,8 @@ def error_n(t, p): errorp : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -198,8 +198,8 @@ def sensitivity(t, p): sensitivity : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -235,8 +235,8 @@ def specificity(t, p): sensitivity : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -272,8 +272,8 @@ def ppv(t, p): PPV : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -309,8 +309,8 @@ def npv(t, p): NPV : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -353,8 +353,8 @@ def mcc(t, p): MCC : float, in range [-1.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.int) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=int) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -372,7 +372,7 @@ def mcc(t, p): if den == 0.0: den = 1.0 - num = np.float((tp*tn)-(fp*fn)) + num = float((tp*tn)-(fp*fn)) return num / den @@ -391,8 +391,8 @@ def auc_wmw(t, p): AUC : float, in range [0.0, 1.0] """ - tarr = np.asarray(t, dtype=np.int) - parr = np.asarray(p, dtype=np.float) + tarr = np.asarray(t, dtype=int) + parr = np.asarray(p, dtype=float) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -426,8 +426,8 @@ def mse(t, p): MSE : float """ - tarr = np.asarray(t, dtype=np.float) - parr = np.asarray(p, dtype=np.float) + tarr = np.asarray(t, dtype=float) + parr = np.asarray(p, dtype=float) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -453,8 +453,8 @@ def r2(t, p): R^2 : float """ - tarr = np.asarray(t, dtype=np.float) - parr = np.asarray(p, dtype=np.float) + tarr = np.asarray(t, dtype=float) + parr = np.asarray(p, dtype=float) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") @@ -480,8 +480,8 @@ def r2_corr(t, p): R^2 : float """ - tarr = np.asarray(t, dtype=np.float) - parr = np.asarray(p, dtype=np.float) + tarr = np.asarray(t, dtype=float) + parr = np.asarray(p, dtype=float) if tarr.shape[0] != parr.shape[0]: raise ValueError("t, p: shape mismatch") --- a/mlpy/ols.py +++ b/mlpy/ols.py @@ -77,8 +77,8 @@ class OLS: target values """ - xarr = np.array(x, dtype=np.float, copy=True) - yarr = np.asarray(y, dtype=np.float) + xarr = np.array(x, dtype=float, copy=True) + yarr = np.asarray(y, dtype=float) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -90,7 +90,7 @@ class OLS: raise ValueError("x, y shape mismatch") xarr = np.concatenate((np.ones((xarr.shape[0], 1), - dtype=np.float), xarr), axis=1) + dtype=float), xarr), axis=1) beta, self._rank = ols_base(xarr, yarr, self._tol) self._beta = beta[1:] @@ -110,7 +110,7 @@ class OLS: if not self._beta or not self._beta0: raise ValueError('no mode computed; run learn() first') - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2 or tarr.ndim < 1: raise ValueError("t must be an 1d or a 2d array_like object") --- a/mlpy/parzen.py +++ b/mlpy/parzen.py @@ -52,8 +52,8 @@ class Parzen: target values """ - K_arr = np.asarray(K, dtype=np.float) - y_arr = np.asarray(y, dtype=np.int) + K_arr = np.asarray(K, dtype=float) + y_arr = np.asarray(y, dtype=int) if K_arr.ndim != 2: raise ValueError("K must be a 2d array_like object") @@ -105,7 +105,7 @@ class Parzen: if self._alpha is None: raise ValueError("no model computed; run learn()") - Kt_arr = np.asarray(Kt, dtype=np.float) + Kt_arr = np.asarray(Kt, dtype=float) if self._kernel is not None: Kt_arr = self._kernel.kernel(Kt_arr, self._x) @@ -115,7 +115,7 @@ class Parzen: raise ValueError("Kt, alpha: shape mismatch") return np.where(s==-1, self._labels[0], self._labels[1]) \ - .astype(np.int) + .astype(int) def alpha(self): """Return alpha. --- a/mlpy/perceptron.py +++ b/mlpy/perceptron.py @@ -57,8 +57,8 @@ class Perceptron: target values (N) """ - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -77,7 +77,7 @@ class Perceptron: ynew = np.where(yarr == self._labels[0], 0, 1) - self._w = np.zeros(xarr.shape[1], dtype=np.float) + self._w = np.zeros(xarr.shape[1], dtype=float) self._bias = 0.0 n = ynew.shape[0] @@ -110,7 +110,7 @@ class Perceptron: if self._w is None: raise ValueError("no model computed") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2: raise ValueError("t must be an 1d or a 2d array_like object") @@ -120,7 +120,7 @@ class Perceptron: raise ValueError("t, model: shape mismatch") return np.where(tmp>0, self._labels[1], self._labels[0]) \ - .astype(np.int) + .astype(int) def w(self): """Returns the coefficients. --- a/mlpy/pls.py +++ b/mlpy/pls.py @@ -51,8 +51,8 @@ class PLS: target values """ - xarr = np.array(x, dtype=np.float, copy=True) - yarr = np.array(y, dtype=np.float, copy=True) + xarr = np.array(x, dtype=float, copy=True) + yarr = np.array(y, dtype=float, copy=True) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -71,9 +71,9 @@ class PLS: xarr = xarr - self._xmean yarr = yarr - self._beta0 - u = np.empty((xarr.shape[1], self._iters), dtype=np.float) - c = np.empty((yarr.shape[1], self._iters), dtype=np.float) - p = np.empty((xarr.shape[1], self._iters), dtype=np.float) + u = np.empty((xarr.shape[1], self._iters), dtype=float) + c = np.empty((yarr.shape[1], self._iters), dtype=float) + p = np.empty((xarr.shape[1], self._iters), dtype=float) for i in range(self._iters): YX = np.dot(yarr.T, xarr) u[:, i] = YX[0] / np.linalg.norm(YX[0]) @@ -111,7 +111,7 @@ class PLS: if self._beta is None: raise ValueError("no model computed; run learn()") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2 or tarr.ndim < 1: raise ValueError("t must be an 1d or a 2d array_like object") --- a/mlpy/rfe.py +++ b/mlpy/rfe.py @@ -23,7 +23,7 @@ from da import KFDAC # used in rfe_kfda def rayleigh(x, kernel, lmb, alpha, idx1, idx2): - R = np.empty(x.shape[1], dtype=np.float) + R = np.empty(x.shape[1], dtype=float) idx = np.ones(x.shape[1], dtype=bool) for i in range(x.shape[1]): idx[i] = False @@ -77,8 +77,8 @@ def rfe_kfda(x, y, p, lmb, kernel): if (p < 0.0) or (p > 1.0): raise ValueError("parameter p must be in [0.0, 1.0]") - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -100,7 +100,7 @@ def rfe_kfda(x, y, p, lmb, kernel): idx2 = np.where(yarr==labels[1])[0] kfda = KFDAC(lmb=lmb, kernel=kernel) - idxglobal = np.arange(xarr.shape[1], dtype=np.int) + idxglobal = np.arange(xarr.shape[1], dtype=int) ranking = [] while True: @@ -158,8 +158,8 @@ def rfe_w2(x, y, p, classifier): if not (hasattr(classifier, 'learn') and hasattr(classifier, 'w')): raise ValueError("parameter classifier must have learn() and w() methods") - xarr = np.asarray(x, dtype=np.float) - yarr = np.asarray(y, dtype=np.int) + xarr = np.asarray(x, dtype=float) + yarr = np.asarray(y, dtype=int) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -174,7 +174,7 @@ def rfe_w2(x, y, p, classifier): if labels.shape[0] != 2: raise ValueError("number of classes must be = 2") - idxglobal = np.arange(xarr.shape[1], dtype=np.int) + idxglobal = np.arange(xarr.shape[1], dtype=int) ranking = [] while True: --- a/mlpy/ridge.py +++ b/mlpy/ridge.py @@ -110,8 +110,8 @@ class Ridge: target values """ - xarr = np.array(x, dtype=np.float, copy=True) - yarr = np.asarray(y, dtype=np.float) + xarr = np.array(x, dtype=float, copy=True) + yarr = np.asarray(y, dtype=float) if xarr.ndim != 2: raise ValueError("x must be a 2d array_like object") @@ -143,7 +143,7 @@ class Ridge: if self._beta is None: raise ValueError("no model computed; run learn()") - tarr = np.asarray(t, dtype=np.float) + tarr = np.asarray(t, dtype=float) if tarr.ndim > 2 or tarr.ndim < 1: raise ValueError("t must be an 1d or a 2d array_like object") @@ -209,8 +209,8 @@ class KernelRidge: target values """ - K_arr = np.asarray(K, dtype=np.float) - y_arr = np.asarray(y, dtype=np.float) + K_arr = np.asarray(K, dtype=float) + y_arr = np.asarray(y, dtype=float) if K_arr.ndim != 2: raise ValueError("K must be a 2d array_like object") @@ -238,7 +238,7 @@ class KernelRidge: # as in G. C. Cawley, N. L. C. Nicola and O. Chapelle. # Estimating Predictive Variances with Kernel Ridge # Regression. - A = np.empty((n+1, n+1), dtype=np.float) + A = np.empty((n+1, n+1), dtype=float) A[:n, :n] = K_arr + self._lmb * np.eye(n) A[n, :n], A[:n, n], A[n, n] = 1., 1., 0. g = np.linalg.solve(A, np.append(y_arr, 0)) @@ -260,7 +260,7 @@ class KernelRidge: if self._alpha is None: raise ValueError("no model computed; run learn()") - Kt_arr = np.asarray(Kt, dtype=np.float) + Kt_arr = np.asarray(Kt, dtype=float) if self._kernel is not None: Kt_arr = self._kernel.kernel(Kt_arr, self._x) --- a/mlpy/stats.py +++ b/mlpy/stats.py @@ -45,7 +45,7 @@ def bootstrap_ci(x, B=1000, alpha=0.05, np.random.seed(seed) - bmean = np.empty(B, dtype=np.float) + bmean = np.empty(B, dtype=float) for b in range(B): idx = np.random.random_integers(0, x_arr.shape[0]-1, x_arr.shape[0]) bmean[b] = np.mean(x_arr[idx]) @@ -65,7 +65,7 @@ def quantile(x, f): percentile `f` should have the value 0.75. """ - xarr = np.array(x, dtype=np.float, copy=True) + xarr = np.array(x, dtype=float, copy=True) xarr = np.ravel(x) xarr.sort() --- a/mlpy/wavelet/padding.py +++ b/mlpy/wavelet/padding.py @@ -70,7 +70,7 @@ def pad(x, method='reflection'): right_x = np.zeros(rdiff, dtype=x_arr.dtype) xp = np.concatenate((left_x, x_arr, right_x)) - orig = np.ones(x_arr.shape[0] + diff, dtype=np.bool) + orig = np.ones(x_arr.shape[0] + diff, dtype=bool) orig[:ldiff] = False orig[-rdiff:] = False --- a/mlpy/fastcluster/fastcluster.py +++ b/mlpy/fastcluster/fastcluster.py @@ -22,7 +22,7 @@ __all__ = ['single', 'complete', 'averag __version_info__ = ('1', '1', '2') __version__ = '.'.join(__version_info__) -from numpy import double, empty, array, ndarray, var, cov, dot, bool, expand_dims, ceil, sqrt +from numpy import double, empty, array, ndarray, var, cov, dot, expand_dims, ceil, sqrt from numpy.linalg import inv from scipy.spatial.distance import pdist