# -*- coding: utf-8 -*-
"""
Created on Wed Nov 30 10:59:16 2022

@author: cpresser
"""

import pandas as pd
import matplotlib.pyplot as plt

#use column 0, Planet, as an index
df = pd.read_csv("PlanetaryFactSheet.csv", index_col=0)

print(df.dtypes)

print(df["Mass"])

#use iloc - access dataframe as a 2d array
print(df.iloc[0, 0])

#loc for conditional matches
print("--------------------------")
#all planets with mas > 100
#produce a data frame of planets matching the criteria
largePlanets = df.loc[ df["Mass"] > 100 ]
print(largePlanets[ ["Mass", "Density"] ])

print("--------------------------")
#use & for and 
# | for or
largeDensePlanets = df.loc[ (df["Mass"] > 100) & (df["Density"] > 1000)]
print(largeDensePlanets)

print("--------------------------")
#get the minimum diameter
print(df.Diameter.min())
print(df["Diameter"].idxmin())


print("--------------------------")
#get the longest day
print(df["Length of Day"].max())
print(df["Length of Day"].idxmax())


print("--------------------------")
#check for correlations
correlate = df.corr()
print(correlate["Distance from Sun"]["Orbital Period"])

print("--------------------------")
df.plot(kind="scatter", x="Orbital Period", y="Distance from Sun")
plt.show()