#!/usr/bin/env python
# coding: utf-8

# In[1]:


from collections import namedtuple
import numpy.matlib as mat
import numpy as np
from numpy.linalg import inv

Match = namedtuple("Match", ["teams", "score"])

def CalcOpr(matches):
    """
    Given a list of matches, to least squares OPR calculation
    """
    num_matches = len(matches) # rows
    teams = list(set().union(*[match.teams for match in matches]))
    num_teams = len(teams)
    
    alliances = mat.zeros((num_matches, num_teams))
    scores = mat.zeros((num_matches, 1))
    
    for idx, match in enumerate(matches):
        scores[idx, 0] = match.score
        for team in match.teams:
            alliances[idx, teams.index(team)] = 1
        
    least_squares_approx = np.dot(
        inv(np.dot(np.transpose(alliances), alliances)),
        np.dot(np.transpose(alliances), scores))
    
    oprs = {}
    for idx, team in enumerate(teams):
        oprs[team] = least_squares_approx[idx, 0] 
    return oprs

print(CalcOpr([
            Match(teams=['A', 'B'], score=10),
            Match(teams=['A', 'C'], score=13),
            Match(teams=['B', 'C'], score=7),
            Match(teams=['A', 'D'], score=15),
            Match(teams=['B', 'D'], score=10),
        ]))


# In[2]:


import tbapy
from pprint import pprint

api_key = "6qOZ9uAEsb4CDrOBNG6ZnIdi9cWBaZ6DHnCSato97Qfo7bBeUwT9NfFt4Gi5sHFN"

tba = tbapy.TBA(api_key)
pprint(tba.status())


# In[3]:


matches = tba.event_matches('2017roe')


# In[4]:


match_outcomes = []
for match in matches:
    match_outcomes.append(
        Match(teams=match['alliances']['blue']['team_keys'],
              score=match['alliances']['blue']['score'])
    )
    match_outcomes.append(
        Match(teams=match['alliances']['red']['team_keys'],
              score=match['alliances']['red']['score'])
    )
score_oprs = sorted(CalcOpr(match_outcomes).items(), key=lambda x: -x[1])
pprint(score_oprs)


# In[5]:


match_outcomes = []
for match in matches:
    match_outcomes.append(
        Match(teams=match['alliances']['blue']['team_keys'],
              score=match['score_breakdown']['blue']['kPaRankingPointAchieved'])
    )
    match_outcomes.append(
        Match(teams=match['alliances']['red']['team_keys'],
              score=match['score_breakdown']['red']['kPaRankingPointAchieved'])
    )
kpa_oprs = CalcOpr(match_outcomes)
pprint(kpa_oprs)


# In[6]:


true_positives = 0
true_negatives = 0
false_positives = 0
false_negatives = 0

for match in matches:
    for alliance in ['red', 'blue']:
        result = match['score_breakdown'][alliance]['kPaRankingPointAchieved']
        prediction = sum(
                kpa_oprs[team] for team in match['alliances'][alliance]['team_keys']
        ) > 0.5
        if result and prediction:
            true_positives += 1
        elif result and not prediction:
            false_negatives += 1
        elif not result and prediction:
            false_positives += 1
        elif not result and not prediction:
            true_negatives += 1

print("True positives", true_positives)
print("True negatives", true_negatives)
print("False positives", false_positives)
print("False negatives", false_negatives)
print("Correctness", (true_positives + true_negatives) / (true_positives + true_negatives + false_positives + false_negatives + 0.0))


# In[7]:


match_outcomes = []
for match in matches:
    match_outcomes.append(
        Match(teams=match['alliances']['blue']['team_keys'],
              score=match['score_breakdown']['blue']['rotorRankingPointAchieved'])
    )
    match_outcomes.append(
        Match(teams=match['alliances']['red']['team_keys'],
              score=match['score_breakdown']['red']['rotorRankingPointAchieved'])
    )
rotor_oprs = CalcOpr(match_outcomes)
pprint(rotor_oprs)


# In[8]:


true_positives = 0
true_negatives = 0
false_positives = 0
false_negatives = 0

for match in matches:
    for alliance in ['red', 'blue']:
        result = match['score_breakdown'][alliance]['rotorRankingPointAchieved']
        prediction = sum(
                rotor_oprs[team] for team in match['alliances'][alliance]['team_keys']
        ) > 0.5
        if result and prediction:
            true_positives += 1
        elif result and not prediction:
            false_negatives += 1
        elif not result and prediction:
            false_positives += 1
        elif not result and not prediction:
            true_negatives += 1

print("True positives", true_positives)
print("True negatives", true_negatives)
print("False positives", false_positives)
print("False negatives", false_negatives)
print("Correctness", (true_positives + true_negatives) / (true_positives + true_negatives + false_positives + false_negatives + 0.0))