Test your AI Agents with the all-new Agent to Agent Testing Platform.Learn More

How to use group_pred method in hypothesis

Best Python code snippet using hypothesis

helpers.py

Source:helpers.py

1import numpy as np2import cvxpy as cp3from sys import argv4import os5from scipy import stats6from scipy.special import softmax7from sklearn.preprocessing import normalize8# generate a numpy matrix of size \R num_samples * m, dist can be "uniform" or "normal"9def generate_data(n_s, m, dist, normalized=False):10    if (dist == 'uniform'):11        X = np.random.uniform(low=0.0, high=1.0, size=(n_s, m))12    elif(dist == 'normal'):13        X = stats.truncnorm.rvs(-1.0, 1.0, 0.0, 1.0, size=(n_s, m))14    15    if normalized:16        norm_X = normalize(X, axis=1, norm='l1')    17        return norm_X18    else:19        return X20# generate matrix L of size d x m according to distribution dist21def generate_loss_matrix(d, m, dist):22    if (dist == 'uniform'):23        L = np.random.uniform(low=0.0, high=1.0, size=(d, m))24    elif(dist == 'normal'):25        L = stats.truncnorm.rvs(-1.0, 1.0, 0.0, 1.0, size=(d, m))26    return L27# generate matrix U of size d x m according to distribution dist28def generate_utility_matrix(d, m, dist):29    if (dist == 'uniform'):30        U = np.random.uniform(low=0.0, high=1.0, size=(d, m))31    elif (dist == 'normal'):32        U = stats.truncnorm.rvs(-1.0, 1.0, 0.0, 1.0, size=(d, m))33    return U34def generate_utility_matrix_var(d, m, dist, var):35    if (dist == 'uniform'):36        U_vec = np.random.uniform(low=0.0, high=1.0, size=(d,))37    elif (dist == 'normal'):38        U_vec = stats.truncnorm.rvs(-1.0, 1.0, 0.0, 1.0, size=(d,))39    40    U_mat = np.ones((d,m))41    for col in range(m):42        noise = np.random.normal(scale=var, size=(d,))43        U_mat[:,col] = np.clip(U_vec+noise, 0, 1)44    45    return U_mat46def define_groups(X, group_dist, normalized=False):47    """ Given a vector where each element represents the portion of the population that should48        belong to each group, and the size of the vector represents the number of groups.49        Return a list of matricies, where each matrix contains all the individuals that belong50        to that group 51    """52    n, m = X.shape53    assert(sum(group_dist)-1 <= 1e-5)54    assert(min(group_dist) >= 0)55    # Let the first element in the feature vector represent group identity56    X0s = X[:,0]57    low = 058    samples_by_group = {}59    for group_id, val in enumerate(group_dist):60        high = low + val61        low_indicies = np.where(X0s >= low)62        high_indicies = np.where(X0s <= high)63        indicies = np.intersect1d(low_indicies, high_indicies)64        65        if normalized:66            norm_X = normalize(X[indicies,:], axis=1, norm='l1')    67            samples_by_group[group_id] = norm_X68        else:69            samples_by_group[group_id] = X[indicies,:]70        low = high71    return samples_by_group72def test_groups():73    X = np.random.uniform(size=(6,6))74    print(X)75    group_dist = [0.25, 0.25, 0.25, 0.25]76    grps = define_groups(X, group_dist)77    print(grps)78def get_default_alpha_arr(K):79    alpha = []80    alpha_value = 1.081    for k in range(K-1):82        alpha_value = alpha_value / 283        alpha.append(alpha_value)84    alpha.append(alpha_value)85    return alpha86def get_optimal_loss(L_mat, X):87    loss_matrix = np.matmul(X, L_mat.T)88    loss_matrix = normalize(loss_matrix, axis=1, norm='l1')89    (n, d) = loss_matrix.shape90    min_val = loss_matrix.min(axis=1)91    assert(min_val.shape == (n,))92    return sum(min_val)/n93    94def predictions(beta_value, all_X):95    return np.argmax(np.matmul(beta_value, all_X.T), axis = 0)96def get_all_predictions(beta_values, all_X, groups, K_):97    learned_predictions = []98    for k in range(K_):99        all_prediction = np.argmax(np.matmul(beta_values[k], all_X.T), axis = 0)100        learned_predictions.append(all_prediction)101    num_groups = len(groups.keys())102    learned_pred_group = {h:[] for h in range(num_groups)} 103    for k in range(K_):104        for h in range(num_groups):105            learned_pred_group[h].append(predictions(beta_values[k], groups[h]))106    107    return learned_predictions, learned_pred_group108def compute_final_loss(alphas, L_mat, X, learned_predictions):109    L_X = np.matmul(X, L_mat.T)110    L_X = normalize(L_X, axis=1, norm='l1')111    112    final_loss = 0113    K = len(alphas)114    n, m = X.shape115    116    for k in range(K):117        for i in range(n):118            final_loss += alphas[k]*L_X[i, learned_predictions[k][i]]119    return final_loss/n120def compute_welfare(alphas, U_mat, X, learned_predictions):121    return compute_final_loss(alphas, U_mat, X, learned_predictions)122def group_utility(alphas, U_mat, group_i, group_j, pred_i, pred_j, same=False):123    ## Compute the utility group i has for group j's clasification124    UX_i = np.matmul(group_i, U_mat.T)125    UX_j = np.matmul(group_j, U_mat.T)126    127    UX_i = normalize(UX_i, axis=1, norm='l1')128    UX_j = normalize(UX_j, axis=1, norm='l1')129    130    ni = group_i.shape[0]131    nj = group_j.shape[0]132    if same:133        welf = compute_welfare(alphas, U_mat, group_i, pred_i)134        return welf135    else:136        welf = 0137        for t, alpha in enumerate(alphas):138            for li in range(ni):139                for lj in range(nj):140                    welf += alpha*UX_i[li, pred_j[t][lj]]141        return welf/(ni*nj) 142def total_group_envy(alphas, U_mat, groups, group_pred):143    violations = 0144    total_envy = 0145    num_groups = len(groups.keys())146    for i in range(num_groups):147        for j in range(num_groups):148            if i != j:149                u_ii = group_utility(alphas, U_mat, groups[i], \150                        groups[i], group_pred[i], group_pred[i], same=True)151                u_ij = group_utility(alphas, U_mat, groups[i], \152                        groups[j], group_pred[i], group_pred[j])    153                total_envy += max(u_ij - u_ii, 0)154                if u_ij > u_ii + 1e-3:155                    violations += 1156    157    total_envy = total_envy / (num_groups*(num_groups-1)) 158    violations = violations / (num_groups*(num_groups-1))159    return total_envy, violations160def total_average_envy(alphas, U_mat, X, pred):161    U_X = np.matmul(X, U_mat.T)162    U_X = normalize(U_X, axis=1, norm='l1')163    n, m = X.shape164    total_envy = 0165    violations = 0166    for i in range(n):167        for j in range(n):168            if i != j:169                u_ii, u_ij = 0, 0170                for k in range(len(alphas)):171                    u_ii += alphas[k]*U_X[i, pred[k][i]]172                    u_ij += alphas[k]*U_X[i, pred[k][j]]173                    total_envy += max(u_ij - u_ii, 0)            174                if u_ij > u_ii + 1e-3:175                    violations += 1176    total_envy = total_envy * (1/(n*(n-1)))177    violations = violations * (1/(n*(n-1)))178    return total_envy, violations179def total_group_equi(alphas, U_mat, groups, group_pred):180    violations = 0181    total_equi = 0182    num_groups = len(groups.keys())183    for i in range(num_groups):184        for j in range(num_groups):185            if i != j:186                u_ii = group_utility(alphas, U_mat, groups[i], \187                        groups[i], group_pred[i], group_pred[i], same=True)188                u_jj = group_utility(alphas, U_mat, groups[j], \189                        groups[j], group_pred[j], group_pred[j], same=True)    190                #print("i: ", i, "j: ", j, "uii:", u_ii, "ujj: ", u_jj) 191                total_equi += abs(u_jj - u_ii)192                if total_equi >= 1e-3:193                    violations += 1194    total_equi = total_equi / (num_groups*(num_groups-1)) 195    violations = violations / (num_groups*(num_groups-1))196    return total_equi, violations197def min_group_welfare(alphas, U_mat, groups, group_pred):198    # return the welfare of group with the lowest welfare199    welfares = []200    num_groups = len(groups.keys())201    for i in range(num_groups):202        u_ii = group_utility(alphas, U_mat, groups[i], \203                    groups[i], group_pred[i], group_pred[i], same=True)204        welfares.append(u_ii)        205    206    return min(welfares)207def get_convex_version(X, Mat_X, Beta, y, i):  208    return cp.max(Mat_X[i, :] + cp.matmul(Beta,X[i,:])) - cp.matmul(Beta[y[i],:], X[i,:])209if __name__ == "__main__":...

translator.py

Source:translator.py

1class Translator(object):2    @staticmethod3    def num_to_string(num):4        if not isinstance(num, int):5            raise TypeError6        elif num < 0:7            raise ValueError8        if num == 0:9            return "zero"10        simple_nums_to_str = {11            1: "one",12            2: "two",13            3: "three",14            4: "four",15            5: "five",16            6: "six",17            7: "seven",18            8: "eight",19            9: "nine",20            10: "ten",21            11: "eleven",22            12: "twelve",23            13: "thirteen",24            14: "fourteen",25            15: "fifteen",26            16: "sixteen",27            17: "seventeen",28            18: "eighteen",29            19: "nineteen"30        }31        if num in simple_nums_to_str:32            return simple_nums_to_str[num]33        second_placed_nums_to_str = {34            2: "twenty",35            3: "thirty",36            4: "forty",37            5: "fifty",38            6: "sixty",39            7: "seventy",40            8: "eighty",41            9: "ninety",42        }43        result_str = ""44        dig_place_groups = []45        while num:46            dig_place_groups.append(num % 100)47            num //= 10048            dig_place_groups.append(num % 10)49            num //= 1050        group_pred = {51            0: "",52            1: " hundred ",53            2: " thousand ",54            4: " million ",55            6: " billion ",56            8: " trillion ",57            10: " quadrillion ",58            12: " quintillion ",59            14: " sextillion ",60            16: " septillion "61        }62        for group, gnum in enumerate(dig_place_groups):63            if gnum == 0:64                continue65            if group % 2 == 0:66                if gnum in simple_nums_to_str:67                    result_str = simple_nums_to_str[gnum] + group_pred[group] + result_str68                else:69                    result_str = second_placed_nums_to_str[gnum // 10]\70                                 + ("-" + simple_nums_to_str[gnum % 10] if gnum % 10 != 0 else "")\71                                 + group_pred[group] + result_str72            else:73                result_str = simple_nums_to_str[gnum] + group_pred[1] + result_str...

Linear_Regression_Classifier.py

Source:Linear_Regression_Classifier.py

1# Linear Regression2import numpy as np3from numpy.linalg import inv4import pickle5import numpy as np6import math7class Linear_Regression_Classifier(object):8  def __init__(self, X, y):9      self.features = X10      self.target = y11  def train(self,l):12      """STUDNET_CODE"""13      x = self.features14      y = self.target15      16      I = np.identity(len(x[1,:]))17      I[0,0] = 018      xT = np.matrix.transpose(x)19      xTx = np.dot(xT,x) + l*I20      XTX = inv(xTx)21      XTX_xT = np.dot(XTX,xT) 22      23      W = XTX_xT.dot(y)24      W = W[np.newaxis]25      Y = np.dot(x,W.T)26      27      self.W = W28      29  def predict(self, X):30      wT = self.W.T31      y = X.dot(wT)32      return y33  34  def encoding_scalar_to_group(self,y):35      group_pred = np.around(y)36      group_pred = np.squeeze(group_pred).T37      return group_pred38  def encdoing_vector_to_group(self,y):39      group_pred = np.squeeze(y)40      group = [np.argmax(r,axis=0) for r in group_pred]41      return group42  def classifier_onehot(self,X):43      y = self.predict(X)44      group_pred=self.encdoing_vector_to_group(y)45      return group_pred46 47  def classifier_scalar(self,X):48      y = self.predict(X)49      group_pred=self.encoding_scalar_to_group(y)50      return group_pred...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.