dataset

dataset.head()

# get the types of each column
dataset.dtypes

#check for missing values
dataset.isnull().sum()

dataset.head()

# create boxplots of each column except % silica concentrate
x = dataset.iloc[:, :-1]
for i in range(0, len(x.columns), 6):
    sns.boxplot(data=x.iloc[:, i:i+6])
    # make the x 
    plt.xticks(rotation=90)
    plt.show()

# creating the correlation matrix
def plotCorrelationMatrix(dataset):
    corr_matrix = dataset.corr()
    # visualizing the correlation matrix
    plt.figure(figsize=(30, 25))  # to set the figure size
    heatmap(corr_matrix, annot=True, lw=0, linecolor='white', cmap='magma', fmt='0.2f')
    plt.title("Correlation Matrix")  # to add title 
    # incrase title size
    plt.title("Correlation Matrix", fontsize=50)
    plt.xticks(rotation=90)
    plt.yticks(rotation=0)
    plt.show()

plotCorrelationMatrix(dataset)
## End

# remove outliers from all columns
x = dataset.iloc[:, :-1]
for i in range(0, len(x.columns)):
    q1 = dataset.iloc[:, i].quantile(0.25)
    q3 = dataset.iloc[:, i].quantile(0.75)
    iqr = q3 - q1
    lower_bound = q1 - (1.5 * iqr)
    upper_bound = q3 + (1.5 * iqr)
    dataset = dataset[(dataset.iloc[:, i] > lower_bound) & (dataset.iloc[:, i] < upper_bound)]

# # create boxplots of each column except % silica concentrate
x = dataset.iloc[:, :-1]
for i in range(0, len(x.columns), 6):
    sns.boxplot(data=x.iloc[:, i:i+6])
    # make the x
    plt.xticks(rotation=90)
    plt.show()

x = dataset.iloc[:, :-1]
corr_features = correlation(x, 0.7)
corr_features

plotCorrelationMatrix(dataset)

dataset.head()

dataset

print(y_train.shape)
print(X_train.shape)

y_train = y_train.ravel()
y_train

print(type(X_train))
print(type(y_train))

from sklearn.linear_model import LinearRegression
regressor = LinearRegression()
regressor.fit(X_train, y_train)

y_test = y_test.ravel()
y_test

print("The Prediction % Silica =",y_pred[6])
print("The Actual % Silica =",y_test[6])

from sklearn.metrics import r2_score, mean_absolute_error

r2score = r2_score(y_test, y_pred)
print("The R2 Score =", r2score * 100, "%")

rmse = np.sqrt(mean_absolute_error(y_test, y_pred))
print("The Root Mean Squared Error =", rmse * 100 ,"%")

y_test = y_test.ravel()
print(type(y_test))
print(y_test.shape)
y_test

print(type(y_pred))
print(y_pred.shape)
y_pred

avg_accuracy = 0
for i in range(len(y_pred)):
    error = abs(y_pred[i] - y_test[i])
    avg_accuracy += 1 - error / y_test[i]


avg_accuracy = avg_accuracy / len(y_pred)
print("The Accuracy of the Model =", round(avg_accuracy, 6) * 100, "%")        

x_predict_test = X_test[45]
y_pred_new = regressor.predict(x_predict_test.reshape(1, -1))
print("The Predicted % Silica =", y_pred_new[0], "%")

from sklearn.linear_model import Lasso
lasso = Lasso()
lasso.fit(X_train, y_train)

print("The Prediction % Silica =",predictions[7], "%")
print("The Actual % Silica =",y_test[7], "%")

r2score = r2_score(y_test, predictions)
print("The R2 Score =", r2score)

error = mean_squared_error(y_test, predictions)
print("The Mean Squared Error =", error)

from sklearn.linear_model import SGDRegressor
sgd = SGDRegressor()
sgd.fit(X_train, y_train)

print("The Prediction % Silica =",predictions[7], "%")
print("The Actual % Silica =",y_test[7], "%")

r2score = r2_score(y_test, predictions)
print("The R2 Score =", r2score)

error = mean_squared_error(y_test, predictions)
print("The Mean Squared Error =", error)