sales=pd.ExcelFile('ElecmartSales.xlsx')
df=sales.parse('Data')
df.head()

df=df[['Date','Day','Time','Region','Card Type','Gender','Buy Category', 'Items Ordered', 'Total Cost', 'High Item']]
df.head()

df.dtypes

df['Date'].dtypes

pd.get_dummies(df['Day'], prefix='Day_of_the_week')

pd.get_dummies(df['Day'], prefix='Day_of_the_week', drop_first=True)

df = pd.concat([df,pd.get_dummies(df['Day'], prefix='Day_of_the_week', drop_first=True)],axis=1)
df.drop(['Day'],axis=1, inplace=True)
df.head()

df = pd.concat([df,pd.get_dummies(df['Time'], prefix='Time_of_the_Day', drop_first=True)],axis=1) #will drop Afternoon
df.drop(['Time'],axis=1, inplace=True)
df = pd.concat([df,pd.get_dummies(df['Region'], prefix='Region', drop_first=True)],axis=1) #will drop MidWest
df.drop(['Region'],axis=1, inplace=True)
df = pd.concat([df,pd.get_dummies(df['Card Type'], prefix='Card', drop_first=True)],axis=1) #will drop Elecmart card
df.drop(['Card Type'],axis=1, inplace=True)
df = pd.concat([df,pd.get_dummies(df['Gender'], prefix='Gender', drop_first=True)],axis=1) #will drop Female
df.drop(['Gender'],axis=1, inplace=True)
df = pd.concat([df,pd.get_dummies(df['Buy Category'], prefix='Buy_Cat', drop_first=True)],axis=1) #will drop High
df.drop(['Buy Category'],axis=1, inplace=True)
df.head()

df_date=df['Date'] #min_max scaler does not work with dates.
##I removed it from the dataframe and saved it so I can add it later if I need it
df.drop(['Date'],axis=1, inplace=True)
df1 = df.values #returns a numpy array
min_max_scaler = preprocessing.MinMaxScaler()
#this scales all the data by column to a value between 0 and 1. you can also scale specific columns only
df_scaled = min_max_scaler.fit_transform(df1)
df = pd.DataFrame(df_scaled) #this process loses the names of the columns
df.head()

data = pd.read_excel('employee_data.xlsx')
data.head()
# copy the data
data_z_scaled = data.copy()
data_z_scaled.head()

column = 'Age'
data_z_scaled[column] = (data_z_scaled[column] - data_z_scaled[column].mean()) / data_z_scaled[column].std()    
  
# view normalized data  
display(data_z_scaled)

column = 'Annual Salary'
data_z_scaled[column] = (data_z_scaled[column] - data_z_scaled[column].mean()) / data_z_scaled[column].std()    
  
# view normalized data  
display(data_z_scaled)

#produce a range of -1 to 1
# copy the data
data_max_scaled = data.copy()
  
# apply normalization techniques on Column 1
column = 'Prior Experience'
data_max_scaled[column] = data_max_scaled[column] /data_max_scaled[column].abs().max()
  
# view normalized data
display(data_max_scaled)

data_copy = data.copy()
train_set = data_copy.sample(frac=0.80, random_state=0) #random state can be omitted, but set value for reproducibility
#frac sets the split proportion: can typically be anywhere from 0.70 to 0.80
test_set = data_copy.drop(train_set.index)

train_set.head()

test_set.head()

regressor = LinearRegression()
regressor.fit(train_set, train_set_labels)

y_pred = regressor.predict(train_set)
y_pred

coeff_df = pd.DataFrame(regressor.coef_, train_set.columns, columns=['Coefficient'])
coeff_df

df_pred = pd.DataFrame({'Actual':train_set_labels, 'Predicted': y_pred})
df_pred.head()

print('Mean Absolute Error:', metrics.mean_absolute_error(train_set_labels, y_pred))  
print('Mean Squared Error:', metrics.mean_squared_error(train_set_labels, y_pred))  
print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(train_set_labels, y_pred)))

print('Mean Absolute Error:', metrics.mean_absolute_error(test_set_labels, y_pred_test))  
print('Mean Squared Error:', metrics.mean_squared_error(test_set_labels, y_pred_test))  
print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(test_set_labels, y_pred_test)))