import pandas
from flask import Flask, render_template
from matplotlib import pyplot as plt
from numpy import vectorize
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error, accuracy_score
from sklearn.model_selection import train_test_split
app = Flask(__name__)
@app.route("/")
def home():
return "" \
"Жукова Алина ПИбд-41
" \
"Лабораторная работа №5
" \
"" \
""
# Регрессия
# 10 вариант - Линейная регрессия
@app.route("/k4_1_task_5", methods=['GET'])
def k4_1_task_5():
data = pandas.read_csv('Data_chess_games.csv')
data = data.loc[data['created_at'] - data['last_move_at'] != 0]
data = data.drop_duplicates()
data = data.sample(n=5000, replace=True, random_state=1)
# отбор нужных столбцов
corr = data[['white_rating', 'black_rating', 'moves', 'opening_eco', 'opening_name', 'winner']]
def SplitPr(stroke):
return len(stroke.split(" "))
corr['len'] = corr['moves'].apply(SplitPr)
corr = corr.loc[corr['len'] >= 10]
number_element = 0
def SelectElemToStolb(elem):
return elem.split(" ")[number_element]
def ScoreXod(elem):
if(elem == "O-O" or elem == "O-O-O"):
return 50
if(elem.__contains__("x")):
return 20
if (elem.__contains__("+")):
return 50
return 10
def EcoToScore(elem):
score_eco = 0
if(str(elem).startswith("A")):
score_eco = 100
if (str(elem).startswith("B")):
score_eco = 200
if (str(elem).startswith("C")):
score_eco = 300
if (str(elem).startswith("D")):
score_eco = 400
if (str(elem).startswith("E")):
score_eco = 500
score_eco += int(str(elem)[1:])
return score_eco
# 0 - white || 1 - black
gamer_ind = 0
def ScoreAllXods(elem):
score = 0
xods = elem.split(" ")
ind_temp = 0
for xod in xods:
if(ind_temp % 2 == gamer_ind % 2):
score += ScoreXod(xod)
ind_temp += 1
return score
corr['score_eco'] = corr['opening_eco'].apply(EcoToScore)
# Ходы
corr['w_score_all'] = corr['moves'].apply(ScoreAllXods)
gamer_ind = 1
corr['b_score_all'] = corr['moves'].apply(ScoreAllXods)
def WinnerSelect(elem):
if(elem == "white"):
return 1
if(elem == "black"):
return -1
return 0
corr['winner'] = corr['winner'].apply(WinnerSelect)
y = corr[['winner']]
# Нормирование к 1
max_zn = corr['white_rating'].max()
min_zn = corr['white_rating'].min()
def normirovanie(elem):
return ((elem - min_zn) / (max_zn - min_zn))
corr['white_rating'] = corr['white_rating'].apply(normirovanie)
max_zn = corr['black_rating'].max()
min_zn = corr['black_rating'].min()
corr['black_rating'] = corr['black_rating'].apply(normirovanie)
corr_st = corr
corr = corr[['white_rating', 'black_rating', 'b_score_all', 'w_score_all', 'score_eco']]
X_train, X_test, y_train, y_test = train_test_split(corr, y, test_size=.05, random_state=42)
linear = LinearRegression()
linear.fit(X_train, y_train)
def FunkPorog(elem):
if(elem >= 0.5):
return 1
if(elem > -0.5):
return 0
if(elem <= -0.5):
return -1
prediction = linear.predict(X_test)
n_predict_vect = vectorize(FunkPorog)
# Средняя ошибка
accuracy_mean_abs_error = mean_absolute_error(prediction, y_test)
# Предсказание исхода партии, в целых числах
predict_acc_score = n_predict_vect(prediction)
accuracy_acc_score = accuracy_score(predict_acc_score, y_test)
white_win = corr_st.loc[corr_st['winner'] == 1]
white_score = (white_win['w_score_all'])
black_score = (white_win['b_score_all'])
white_win_x = white_score - black_score + white_win['score_eco']
white_win_y = white_win['winner'] + (white_win['black_rating'] - white_win['white_rating'])
black_win = corr_st.loc[corr_st['winner'] == -1]
white_score = (black_win['w_score_all'])
black_score = (black_win['b_score_all'])
black_win_x = white_score - black_score + black_win['score_eco']
black_win_y = black_win['winner'] + (black_win['black_rating'] - black_win['white_rating'])
no_win = corr_st.loc[corr_st['winner'] == 0]
white_score = (no_win['w_score_all'])
black_score = (no_win['b_score_all'])
no_win_x = white_score - black_score + no_win['score_eco']
no_win_y = no_win['winner'] + (no_win['black_rating'] - no_win['white_rating'])
white_win = white_win[['white_rating', 'black_rating', 'b_score_all', 'w_score_all', 'score_eco']]
black_win = black_win[['white_rating', 'black_rating', 'b_score_all', 'w_score_all', 'score_eco']]
no_win = no_win[['white_rating', 'black_rating', 'b_score_all', 'w_score_all', 'score_eco']]
plt.subplot(1, 2, 1)
plt.scatter(white_win_x, (corr_st.loc[corr_st['winner'] == 1])['winner'], alpha=0.6, s=25, c='red')
plt.scatter(black_win_x, (corr_st.loc[corr_st['winner'] == -1])['winner'], alpha=0.6, s=25, c='blue')
plt.scatter(no_win_x, (corr_st.loc[corr_st['winner'] == 0])['winner'], alpha=0.6, s=25, c='green')
plt.scatter(white_win_x, white_win_y, alpha=0.2, s=25, c='pink')
plt.scatter(black_win_x, black_win_y, alpha=0.2, s=25, c='cyan')
plt.scatter(no_win_x, no_win_y, alpha=0.2, s=25, c='yellow')
plt.savefig('static\k4_1_5_regression.png')
return "" \
"Регрессия
" \
"Вариант 10. Задание 5 - Линейная регрессия
" \
" Точность модели:
" \
" Отклонение по абсолютной ошибке: " + str(accuracy_mean_abs_error) + "
" \
" Отклонение по исходу партии в целых числах: " + str(accuracy_acc_score) + "
" \
"" + render_template(
'4_1_l6_figure1.html') + "
" \
""
if __name__ == "__main__":
app.run(debug=True)