def working_random_forest():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#Regression with random forrest
from sklearn.ensemble import RandomForestRegressor
regressor = RandomForestRegressor(n_estimators=300,
criterion="mse",
random_state=0)
regressor.fit(MLobj.X, MLobj.y)
#Predict
y_pred = regressor.predict(6.5)
print(y_pred)
#Visualize
import matplotlib.pyplot as plt
X_grid = np.arange(min(MLobj.X), max(MLobj.X), 0.01)
X_grid = X_grid.reshape((len(X_grid), 1))
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(X_grid, regressor.predict(X_grid), color="blue")
plt.title("Truth or Bluff (SVR)")
plt.xlabel("Position Salary")
plt.ylabel("Salary")
plt.show()
def working_dec_reg_tree():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#Regression with decission regression tree
from sklearn.tree import DecisionTreeRegressor
regressor = DecisionTreeRegressor(random_state=0)
regressor.fit(MLobj.X, MLobj.y)
#Predict
y_pred = regressor.predict(6.5)
print(y_pred)
#Visualize
import matplotlib.pyplot as plt
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(MLobj.X, regressor.predict(MLobj.X), color="blue")
plt.title("Truth or Bluff (SVR)")
plt.xlabel("Position Salary")
plt.ylabel("Salary")
plt.show()
X_grid = np.arange(min(MLobj.X), max(MLobj.X), 0.01)
X_grid = X_grid.reshape((len(X_grid), 1))
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(X_grid, regressor.predict(X_grid), color="blue")
plt.title("Truth or Bluff (SVR)")
plt.xlabel("Position Salary")
plt.ylabel("Salary")
plt.show()
def preco2():
#Preping data
#We create the EasyReg object to read file and explore data.
MLobj = EasyReg()
MLobj.read(
"C:/Users/4256GU/Desktop/Data Analytics/PRECO2/donnees/inputs_clean_csv 2.csv",
encod='cp1252',
delim=";")
MLobj.explore()
#Split X and Y. We'll predict the gaz consomption (Y)
MLobj.split_X_y(1)
#We keep with X the current meteo, previous gaz consumption, vacations, labor day and previous temperatures.
#MLobj.X=MLobj.X[:,[1]+range(7,31)+[31,32]+range(33,57)+range(64,88)]
MLobj.X = MLobj.X[:, [1] + list(range(7, 31)) + [31, 32] +
list(range(33, 57)) + list(range(64, 88))]
#See what we get at X
print(MLobj.myDS.columns[[1] + list(range(7, 31)) + [31, 32] +
list(range(33, 57)) + list(range(64, 88))])
MayML.exploreArray(MLobj.X, "X")
def working_LR_Salary():
MLobj = EasyReg()
MLobj.read("Salary_Data.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.split_ds(ts=1 / 3)
MLobj.fitLR()
y_pred = MLobj.predict()
MLobj.visualize_testingDS_vs_pred()
MLobj.visualize_trainingDS_vs_pred()
def working_MR_Startups():
MLobj = EasyReg()
MLobj.read("50_Startups.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.encode_categorial_dummy_X([3])
MLobj.split_ds(test_set=0.2)
MLobj.fitLR()
y_pred = MLobj.predict()
def working_model_bwrd_elimination():
#Prepare data
MLobj = EasyReg()
MLobj.read("50_Startups.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.encode_categorial_dummy_X([3])
MLobj.split_ds(test_set=0.2)
#We delete a dummy variable, but we don't actually need it
MLobj.X = MLobj.X[:, 1:]
#Append interceptor
MLobj.X = np.append(arr=np.ones((50, 1)).astype(int),
values=MLobj.X,
axis=1)
#Backwards elimination
#1st round
X_opt = MLobj.X[:, [0, 1, 2, 3, 4, 5]]
regressor_OLS = sm.OLS(endog=MLobj.y, exog=X_opt).fit()
#Look for highest P-value feature
regressor_OLS.summary()
#2nd round
X_opt = MLobj.X[:, [0, 1, 3, 4, 5]]
regressor_OLS = sm.OLS(endog=MLobj.y, exog=X_opt).fit()
#Look for highest P-value feature
regressor_OLS.summary()
#3rd round
X_opt = MLobj.X[:, [0, 3, 4, 5]]
regressor_OLS = sm.OLS(endog=MLobj.y, exog=X_opt).fit()
#Look for highest P-value feature
regressor_OLS.summary()
#4rth round
X_opt = MLobj.X[:, [0, 3, 5]]
regressor_OLS = sm.OLS(endog=MLobj.y, exog=X_opt).fit()
#Look for highest P-value feature
regressor_OLS.summary()
#5th round
X_opt = MLobj.X[:, [0, 3]]
regressor_OLS = sm.OLS(endog=MLobj.y, exog=X_opt).fit()
#Look for highest P-value feature
regressor_OLS.summary()
def working_build_bwrd_elimination():
#Prepare data
MLobj = EasyReg()
MLobj.read("50_Startups.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.encode_categorial_dummy_X([3])
#We delete a dummy variable, but we don't actually need it
MLobj.X = MLobj.X[:, 1:]
#Add interception
MLobj.append_interceptor()
SL = 0.05
X_opt = MLobj.X[:, [0, 1, 2, 3, 4, 5]]
#X_Modeled = MLobj.backwardElimination_Pvalue(X_opt,MLobj.y, SL)
MLobj.backwardElimination(X_opt, MLobj.y, SL, inPlace=True)
MLobj.split_ds(test_set=0.2)
MLobj.fitLR()
y_pred = MLobj.predict()
print(y_pred)
def debug_PR():
#testing
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#MLobj.encode_categorial_dummy_X([0])
MLobj.split_ds(test_set=0)
from sklearn.preprocessing import PolynomialFeatures
poly_reg = PolynomialFeatures(degree=4)
X_poly = poly_reg.fit_transform(MLobj.X)
poly_reg.fit(X_poly, MLobj.y)
from sklearn.linear_model import LinearRegression
lin_reg_2 = LinearRegression()
lin_reg_2.fit(X_poly, MLobj.y)
import matplotlib.pyplot as plt
X_grid = np.arange(min(MLobj.X), max(MLobj.X), 0.1)
X_grid = X_grid.reshape((len(X_grid), 1))
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(X_grid,
lin_reg_2.predict(poly_reg.fit_transform(X_grid)),
color="blue")
plt.title("test")
plt.xlabel("pos label")
plt.ylabel("salary")
plt.show()
def working_polynomial_regression():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#MLobj.encode_categorial_dummy_X([0])
MLobj.split_ds(test_set=0)
#Fit LR, predict and visualize
MLobj.fitPR()
MLobj.predict()
MLobj.visualize_trainingDS_vs_pred()
print(MLobj.predictVar(6.5))
#Fit PR, predict and visualize
MLobj.fitPR(2)
MLobj.predict(PR=True)
MLobj.visualize_trainingDS_vs_pred(PR=True)
#Fit PR, predict and visualize
MLobj.fitPR(3)
MLobj.predict(PR=True)
MLobj.visualize_trainingDS_vs_pred(PR=True)
#Fit PR, predict and visualize
MLobj.fitPR(4)
MLobj.predict(PR=True)
MLobj.visualize_trainingDS_vs_pred(PR=True)
#Fit PR, predict and visualize, more granularity
MLobj.fitPR(4)
MLobj.predict(PR=True)
x_grid = MLobj.sample_change_resolution(sampleX=MLobj.X, gran=0.1)
MLobj.visualize_trainingDS_vs_pred(PR=True, xsample=x_grid)
print(MLobj.predictVar(6.5, PR=True))
def debug_predict():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#MLobj.encode_categorial_dummy_X([0])
MLobj.split_ds(test_set=0)
from sklearn.linear_model import LinearRegression
lin_reg = LinearRegression()
lin_reg.fit(MLobj.X, MLobj.y)
lin_reg.predict(6.5)
def working_SVR():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#Feature scaling
from sklearn.preprocessing import StandardScaler
sc_X = StandardScaler()
sc_y = StandardScaler()
MLobj.X = sc_X.fit_transform(MLobj.X)
MLobj.y = sc_y.fit_transform(MLobj.y.reshape(-1, 1))
from sklearn.svm import SVR
regressor = SVR(kernel="rbf")
regressor.fit(MLobj.X, MLobj.y)
y_pred = regressor.predict(sc_X.transform(np.array([[6.5]])))
pred = sc_y.inverse_transform(y_pred)
import matplotlib.pyplot as plt
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(MLobj.X, regressor.predict(MLobj.X), color="blue")
plt.title("Truth or Bluff (SVR)")
plt.xlabel("Position Salary")
plt.ylabel("Salary")
plt.show()
X_grid = np.arange(min(MLobj.X), max(MLobj.X), 0.1)
X_grid = X_grid.reshape((len(X_grid), 1))
plt.scatter(MLobj.X, MLobj.y, color="red")
plt.plot(X_grid, regressor.predict(X_grid), color="blue")
plt.title("Truth or Bluff (SVR)")
plt.xlabel("Position Salary")
plt.ylabel("Salary")
plt.show()
def working_SVR_easy():
#Prepare data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
#Preparing data
MLobj.explore()
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
#Feature scaling
MLobj.split_ds(test_set=0)
MLobj.scale_features(scaleY=True)
#Fit model
MLobj.fitSVR()
#Predict
y_pred = MLobj.predictVar(6.5)
print("prediction : ", y_pred, "\n")
#Visualize
MLobj.visualize_trainingDS_vs_pred()
X_grid = np.arange(min(MLobj.X_train), max(MLobj.X_train), 0.1)
X_grid = X_grid.reshape((len(X_grid), 1))
MLobj.visualize_trainingDS_vs_pred(X_grid)
def working_drt_easy():
#Read data
MLobj = EasyReg()
MLobj.read("Position_Salaries.csv")
#Check data
MLobj.explore()
#Prepare data
MLobj.split_X_y()
MLobj.X = MLobj.X[:, 1:2]
MLobj.split_ds(test_set=0)
#Regression with decission regression tree
MLobj.fitDRT()
#Predict
y_pred = MLobj.predictVar(6.5)
print("prediction : ", y_pred, "\n")
#Visualize
MLobj.visualize_trainingDS_vs_pred()
X_grid = np.arange(min(MLobj.X_train), max(MLobj.X_train), 0.1)
X_grid = X_grid.reshape((len(X_grid), 1))
MLobj.visualize_trainingDS_vs_pred(X_grid)