#import vital packages for proper analysis
import time
import numpy as np
from matplotlib import pyplot as plt
import seaborn as sns
from pandas import Series, DataFrame
import pandas as pd
from scipy import stats
%matplotlib inline

# Read the Police Shooting data set
data_df=pd.read_csv('/home/mint/fatal-police-shootings-data.csv')

# Print the shape & Head of the data frame
print(data_df.shape)
print(data_df.head(6))

#Properly assigning categorical records as a category
data_df.id = data_df.id.astype('category')
data_df.armed = data_df.armed.astype('category')
data_df.gender = data_df.gender.astype('category')
data_df.city = data_df.city.astype('category')
data_df.state = data_df.state.astype('category')
data_df.race = data_df.race.astype('category')
data_df.threat_level = data_df.threat_level.astype('category')
data_df.flee = data_df.flee.astype('category')
data_df.manner_of_death = data_df.manner_of_death.astype('category')

#Properly naming each one of the races, to facilitate analysis and comprehension in visualizations
data_df.replace(to_replace = ['A'], value = ['Asian'], inplace = True)
data_df.replace(to_replace = ['B'], value = ['Black'], inplace = True)
data_df.replace(to_replace = ['H'], value = ['Hispanic'], inplace = True)
data_df.replace(to_replace = ['N'], value = ['Native American'], inplace = True)
data_df.replace(to_replace = ['O'], value = ['Other'], inplace = True)
data_df.replace(to_replace = ['W'], value = ['White'], inplace = True)

data_df['month'] = pd.to_datetime(data_df['date']).dt.month
data_df['year'] = pd.to_datetime(data_df['date']).dt.year

data_df.head()

#The majority of crimes are committed by 3 racial groups. White, Black and Hispanic
df3.race.value_counts(normalize=True).plot(kind='pie', figsize = (8,8))
plt.title('Deaths by Race\nNormalized Data')

# Top States
plt.xlabel("Frequency")
plt.ylabel("City")
plt.title("Top 10 States with Most Fatal Police Shootings")
plt.barh(df3[STATE].value_counts(normalize=True)[:10].index, df3[STATE].value_counts()[:10].values)

# Sum by percentage
# we can see that the top 10 states in the US account for 53.32% 
# of all deaths in the US. Might be worth focusing on these states to look for trends
df3.state.value_counts(normalize=True)[:10].sum()

# Listing Specifically Black vs. White
RaceList = ['White', 'Black']
df3_race = df3[df3.race.isin(RaceList)]
df3_race.race.unique()

CityList = ['Los Angeles','Phoenix','Houston','Las Vegas','San Antonio','Columbus','Chicago','Albuquerque','Kansas City','Jacksonville']
df3_race_city = df3_race[df3_race.city.isin(CityList)]
df3_race_city.city.unique()

df3_race_city.groupby('race').city.value_counts(normalize=True).unstack().plot(kind='bar', figsize=(18,8))
plt.title('Deaths Per Race and City')
plt.ylabel('% of Total Deaths per Race')

# Adding a correlation (corrmat) as well as heatmap
corrmat = data_df.corr()
print(corrmat)

#sns.heatmap (corrmat, vmax=.8, square=True)
sns.set()
f, ax = plt.subplots(figsize=(12,8))
heatmap=data_df.corr()
sns.heatmap(heatmap, vmax=.8, square=True, annot=True)

data_df.race.unique()

# Here I have added two columns to the df3 dataframe named - white and non_white.
# Next, I map and convert the race for white as 1 and non-white as 0 
# finally, I display the head of the df3 dataframe o confirm the mapping.

df3['white'] = 'white'
df3['non_white'] = 'non_white'
df3['white'] = df3.race.map ({'Asian':0,'White':1,'Hispanic':0,'Black':0,'Other':0,'Native American':0})
df3['non_white'] = df3.race.map ({'Asian':1,'White':0,'Hispanic':1,'Black':1,'Other':1,'Native American':1})

df3.dropna(inplace=True)
df3.head()

# Disply the scatter plot of Age v. Non -white deaths from the updated df3 dataframe
x = df3['age']
y = df3['non_white'] 

logR = LogisticRegression()
plt.scatter(x,y, marker='+',color='red')
plt.xlabel("Age")
plt.ylabel("Non-White Deaths")

# I believe the result of the Linear Regression Model should really be a Logistic Regression Model.
slope, intercept, r, p, std_err = stats.linregress(x, y)

def myfunc(x):
  return slope * x + intercept

mymodel = list(map(myfunc, x))

plt.scatter(x, y, marker='+',color='red')
plt.xlabel("Age")
plt.ylabel("Non-White Deaths")
plt.plot(x, mymodel)

X_test

X_train

model.fit(X_train,y_train)

model.predict(X_test)

# Accuancy of my model prediction base on X_Test
model.score(X_test,y_test)

model.predict_proba(X_test)