#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sat Jul  1 12:05:42 2023

@author: richardson
"""

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# dataframes and histograms

# new way to construct a dataframe
# uses a "dictionary"
df = pd.DataFrame(
    {
     'S': np.random.randn(1000)*3+2,
     # mean =2, stdev=3
     "T": np.random.randn(1000),
     # mean 0, stddev 1
     "U": np.random.randn(1000)-4
     # mean -4, stddev 1
     }, columns = list('STU')
    )
#print(df.head())
#print(df.info())

#plt.figure()
#df.plot.hist(bins=40
  #          ,alpha=0.5)
#plt.show()

df2 = pd.DataFrame(np.random.rand(50,4),
                   columns=['a','b','c','d'])
#df2.plot.scatter(x='a',y='b')
#df2.plot.scatter(x='a',y='b',c='c',s=df2['d']*100)

# making a series data frame
df3 = pd.Series({1:1,4:8,9:27,16:64,17:17})
print(df3)

bubba = [[1,2,3],[-2,3,-7],[4,4,4],[0,0,5]]
cols = ['hi','there','yall']
df4 = pd.DataFrame(data=bubba,columns=cols)
print(df4)
#print(df4[2][1]) does not work
print(df4.loc[2,'there'])
print(df4.at[2,'there'])
print(df4.iloc[[2],[1]])
print(df4.iat[2,1])

print(df4.loc[2]) #row with index 2
print(df4.loc[:,'there']) # column 'there'