# -*- coding: utf-8 -*-
"""
Created on Tue Mar 22 15:16:43 2022

@author: Sannidhya 
"""

""" Assignment 2 """

import pandas as pd
import numpy as np
import seaborn as sb
import matplotlib as plt
# Q1
prob1 = pd.read_csv(r"C:\Users\Home\Desktop\Python Files\p1.csv")

prob1.isna().sum()

job = prob1[prob1["Location"].astype("str").str.contains("Mumbai")]

# There are 3731 jobs in Mumbai

metros = ["Bengaluru", "Kolkata", "Delhi", "Chennai", "Gurgaon", "Pune"]

m_jobs = 0
for i in metros:
    m_jobs = m_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
m_jobs

list = ["Bengaluru", "Kolkata", "Delhi", "Chennai", "Gurgaon", "Pune"]
bengaluru_jobs = 0
kolkata_jobs = 0
delhi_jobs = 0
chennai_jobs = 0
gurgaon_jobs = 0
pune_jobs = 0
for i in list:
    if (i == "Bengaluru"): 
        bengaluru_jobs = bengaluru_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 5529
    if (i == "Kolkata"): 
        kolkata_jobs = kolkata_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 373
    if (i == "Delhi"): 
        delhi_jobs = delhi_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 1834      
    if (i == "Chennai"): 
        chennai_jobs = chennai_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 1518
    if (i == "Gurgaon"): 
        gurgaon_jobs = gurgaon_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 2770
    if (i == "Pune"): 
        pune_jobs = pune_jobs + prob1[prob1["Location"].astype("str").str.contains(i)].shape[0]
  # 1781
 
# Q2

sb.displot(prob1["Experience"], kind = "hist")        
sb.histplot(prob1["Experience"], kde = True)


opening = pd.value_counts(prob1.Experience).reset_index()
opening.columns = ["Experience", "Frequency"]
top5 = opening.iloc[0:5,:]
import matplotlib.pyplot as plt
plt.bar(top5.Experience, top5.Frequency, color = "cyan")


fresher_jobs = opening.iloc[0:6, :]
total_fresher_jobs = sum(fresher_jobs.Frequency)