import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from IPython.html.widgets import interact
import matplotlib.mlab as mlab
import seaborn as sb
#sb.set()  # Reset Style
sb.set(style="white", palette="muted")
sb.set_context("talk")

%matplotlib inline

df = pd.read_csv('2014-06-27-001-Data.csv')

# Convert Epoch Millis to Timestamp
# http://pandas.pydata.org/pandas-docs/dev/timeseries.html#epoch-timestamps
df.index = pd.to_datetime(df['millis'], unit='ms')

accelerations = df[['ax','ay']].dropna()

accelerations.plot();

import scipy.stats as stats
from IPython.html.widgets import interact
dists = ['Normal', 'Rayleigh', 'Weibull']

@interact
def plot_sb_dist(column=accelerations.columns.tolist(), dist=dists):
    plt.figure(figsize=(10, 4))
    dist_map = {
        'Rayleigh': stats.rayleigh,
        'Weibull': stats.exponweib,
        'Normal': stats.norm,
    }
    sb.distplot(accelerations[column], fit=dist_map[dist], label='# of Sensor Readings')
    sb.plt.legend()
    plt.savefig("ax_dist.png", dpi=72, bbox_inches='tight')

accelerations.describe()

gps = pd.DataFrame()

arcLon = 2*np.pi*6378.0/360.0
arcLat = arcLon * np.cos(df.longitude*np.pi/180.0)

gps['LonM'] = df.longitude * arcLon * 1000.0
gps['LatM'] = df.latitude * arcLat * 1000.0

sb.jointplot(gps.LonM, gps.LatM, kind='kde', \
             xlim=(gps.LonM.min()-1, gps.LonM.max()+1), \
             ylim=(gps.LatM.min()-1, gps.LatM.max()+1), \
             size=10, space=1)
plt.savefig('gps_dist.png', dpi=150)

@interact
def plot_sb_dist(column=gps.columns.tolist(), dist=dists):
    plt.figure(figsize=(10, 6))
    dist_map = {
        'Rayleigh': stats.rayleigh,
        'Weibull': stats.exponweib,
        'Normal': stats.norm,
    }
    sb.distplot(gps[column], fit=dist_map[dist])

gps.describe()

def pltnormpdf(mean, variance):
    plt.figure(figsize=(8,5))
    plt.plot(x,mlab.normpdf(x, mean, variance))
    plt.ylim(0,1)

x = np.linspace(-10,10,500)
interact(pltnormpdf, mean=(-5,5,0.5), variance=(0.1,10,0.1));