VincentTatan

def create_rfe_feature_selection(df, target_feature, n_features=3):
  columns_list = list(df.columns)
  columns_list.remove(target_feature)

  logreg = LogisticRegression(solver='lbfgs')
  rfe = RFE(logreg, n_features)
  rfe = rfe.fit(df[columns_list],df[target_feature].values.ravel())
  filtered_rfe_list = list(compress(df.columns, rfe.support_))
  print(rfe.support_)
  print(rfe.ranking_)

VincentTatan

def create_dummies_from_categorical_column (df,categorical_features):
  df_copy = df.copy()
  print('changing {}'.format(list(categorical_features)))
  df_result = pd.DataFrame()
  for feature in categorical_features:
    df_temp = df_copy[feature].str.get_dummies()
    df_temp.columns = ['{}_{}'.format(feature,column) for column in df_temp.columns]
    df_copy = df_copy.join(df_temp,rsuffix='_dept')
  df_copy.drop(categorical_features,axis=1,inplace=True)  
  return df_copy

VincentTatan

def stack_plot(title, ylabel, dates, val_pairs, ymax=None, legend=None):
  # get series values
  fig, ax = plt.subplots(1,1)
  fig.set_figwidth(20)
  fig.set_figheight(10)
  ax.set_title(title, size=20)
  ax.set_ylabel(ylabel, size=20)
  ax.set_xlim((0, len(dates)))
  ax.minorticks_on()
  ax.grid(which='major', axis='y', linestyle='-', linewidth=2, color='k')

VincentTatan

sns.pairplot(iris, hue="species")

VincentTatan

def create_box_and_scatter_plot(df,feature_1,feature_2,target_feature,figsize=(10,5)):
  fig, ax =plt.subplots(1,2,figsize=figsize)
  sns.boxplot(x=target_feature, y= feature_1, data=df, ax=ax[0]).set_title(feature_1)
  sns.boxplot(x=target_feature, y= feature_2, data=df, ax=ax[1]).set_title(feature_2)
  fig2, ax2 =plt.subplots(1,1,figsize=figsize)
  sns.scatterplot(x=feature_1, y=feature_2, hue=target_feature, data=df, ax = ax2)

VincentTatan

def create_heatmap(df,figsize=(10,10)):
  corr = df.corr()
  fig, ax = plt.subplots(figsize=figsize)
  ax.set_title("Heatmap")
  sns.heatmap(corr,ax=ax, annot=True, linewidths=.5)

VincentTatan

sns.factorplot(x='sales',data=salary_df,col='left',kind='count',aspect=.8,size=10)

VincentTatan

def create_layered_boxplot(df,features, target_feature, subtitle='layered boxplot', kind='count',figsize=(20,7)):
  n_charts = len(features)
  fig, axes = plt.subplots(ncols=n_charts,figsize=figsize)
  fig.suptitle(subtitle, fontsize=16)
  for i in range(n_charts):
    feature = features[i-1]
    sns.boxplot(x=target_feature,y=feature,data=df,ax=axes[i-1])
  df[feature].describe()

VincentTatan

def draw_plot(df,features,subtitle,figsize=(20,3),type='distplot'):
  n_charts = len(features)
  fig, axes = plt.subplots(ncols=n_charts,figsize=figsize)
  fig.suptitle(subtitle, fontsize=16)
  for i in range(n_charts):
    feature = features[i-1]
    
    if len(features) >1:
      ax=axes[i-1]
    else:

VincentTatan

def create_pie(df,target_variable,figsize=(10,10)):
  print(df[target_variable].value_counts())
  fig, ax = plt.subplots(figsize=figsize)
  ax.pie(df[target_variable].value_counts().values,  labels=df[target_variable].value_counts().index,autopct = '%1.2f%%',textprops={'fontsize': 20})
  ax.axis('equal')
  plt.title(target_variable)
  plt.show()