PANDAS TRICKS

pandas_tricks.py

# Split and expand string values into columns 

df[‘name’] = df.name.str.split(" ", expand=True)

# GRoup by two columns

gkk = df.groupby(['Team', 'Position']) 


# Merge two df with common columns

#df_values = pd.concat([df_com_count,max_work_com ], axis=1)
df_values = pd.merge(df_com_count,max_work_com, on = "Ministerio",  how='outer')
df_values.fillna(0, inplace=True)
df_values.rename(columns={
    "TotalWork_COM": "Acciones: Comunicacion",
    "TotalWork": "Acciones: Matriz de Acciones"
}, inplace=True)

df_values.info()


# SNS BAR PLOT WITH ROTATIONS

def plot_bar(df_col, title  = 'Call-center, top 10 coductas'):
    
    city_count  = df_col.value_counts()
    
    city_count = city_count[:10,]
    plt.figure(figsize=(10,5))

    plot_bar_count = sns.barplot(city_count.index, city_count.values, alpha=0.8)

    plt.setp(plot_bar_count.get_xticklabels(), rotation=90)

    plt.title(title)

    plt.ylabel('Number of Calls', fontsize=12)
    plt.xlabel('Tipo de Conducta', fontsize=12)
    plt.show()
    

def create_timestamp_indexes(self, df, column="", format=""):
    """
    Create datestamp index to dataframe
    """

    df['time'] =  pd.to_datetime(df[column], format=format)

    # df = df.drop(columns_to_remove, 1)
    df.set_index('time', inplace=True) 

    return df


# Create excel sheets

# Create a Pandas Excel writer using XlsxWriter as the engine.
writer = pd.ExcelWriter('pandas_multiple.xlsx', engine='xlsxwriter')

# Write each dataframe to a different worksheet.
df1.to_excel(writer, sheet_name='Sheet1')
df2.to_excel(writer, sheet_name='Sheet2')
df3.to_excel(writer, sheet_name='Sheet3')

# Close the Pandas Excel writer and output the Excel file.
writer.save()
  
# GRoup by and count


max_work = remote_df.groupby("Ministerio")["Tareas"].agg(np.max).sort_values(ascending=True).reset_index(name="TotalWork")


# REGEX FIND BETWEEN:

import re

# SEARCH BETWEEN PIPES
pattern = re.compile(r'(?<=\?id=)[^&comment_id]*')

# Iterar sobre la lista y buscar lo que esta dentro del patron de pipes | ... |

match = re.search(pattern, text )

if match:
    name = match.group(0)
    print(name)
    
    
#############################


df_comments_top_N00 = df_comments.assign(
                        topN00 = df_comments["commenter"].isin(df_comments_topN00_c["commenter"]).astype(int)
                    )


###############################
# Parsear la fecha datetime

from dateutil import parser

df_comments["timestamp"] = df_comments["comment_timestamp"].apply(lambda x: parser.parse(x) )

print(df_comments.shape)
df_comments.head()






# Barras del numero de post que se tiene de las paginas scrapeadas

df_post_counts = df_posts_raw["real_page_name"].value_counts()

ax = df_post_counts.plot(kind="barh", figsize=(13,8))

for i, v in enumerate(df_post_counts):
    ax.text(v + 1, i + -.15, str(v), color='blue', fontweight='bold')