def get_x_matrix(var_to_calc, restricted=False):
var_to_calc -= 1
x_2 = read_column_from_csv(column_number=0 + var_to_calc * 4,
file='data/4problem.csv')
x_3 = read_column_from_csv(column_number=1 + var_to_calc * 4,
file='data/4problem.csv')
x_4 = read_column_from_csv(column_number=2 + var_to_calc * 4,
file='data/4problem.csv')
y_1 = read_column_from_csv(column_number=3 + var_to_calc * 4,
file='data/4problem.csv')
len_of_data = len(y_1)
# vector of MSE coefficients is (X^T * X)^-1 * X^T * Y
# 1. make vector-column
x = numpy.ones((len_of_data, 1), dtype=float)
x_2 = numpy.array([x_2]).T
x_3 = numpy.array([x_3]).T
x_4 = numpy.array([x_4]).T
y_1 = numpy.array([y_1]).T
x = numpy.concatenate((x, x_2), axis=1)
if not restricted:
x = numpy.concatenate((x, x_3), axis=1)
x = numpy.concatenate((x, x_4), axis=1)
return x
def ls(var_to_calc, ridge=0, restricted=False):
"""
:param ridge: ridge coefficient
:param var_to_calc: variant from 1 to 10
:param restricted: model with restriction or not
:return: array of LS(least squares) coefficients
"""
x = get_x_matrix(var_to_calc, restricted)
# (X^T * X)
# numpy.dot is a matrix multiplication
# if ridge is none 0
if not restricted:
ar = numpy.zeros((4, 4), float)
numpy.fill_diagonal(ar, float(ridge))
x_step1 = numpy.dot(x.T, x) + ar
else:
x_step1 = numpy.dot(x.T, x)
# (X^T * X)^-1
x_step2 = linal.inv(x_step1)
# (X^T * X)^-1 * X^T
x_step3 = numpy.dot(x_step2, x.T)
# (X^T * X)^-1 * X^T * Y
y_1 = read_column_from_csv(column_number=3 + (var_to_calc - 1) * 4,
file='data/4problem.csv')
y_1 = numpy.array([y_1]).T
coefficient_vector = numpy.dot(x_step3, y_1)
return coefficient_vector.T[0]
for i in range(len(prev_centers)):
prev_centers[i] = list(centers[i]) # we must copy this way.
for i in range(len(centers)): # calculating new centers
sx = 0
sy = 0
for j in range(len(clusters[i])):
sx += clusters[i][j][0]
sy += clusters[i][j][1]
if len(clusters[i]
) > 0: # situation when cluster is empty is possible
centers[i][0] = round(sx / len(clusters[i]), 5)
centers[i][1] = round(sy / len(clusters[i]), 5)
is_continue = False
for i in range(len(centers)): # decide must we continue or not
if centers[i] not in prev_centers:
is_continue = True
break
return clusters
X = read_column_from_csv(0 + (v_number - 1) * 2, 'data/7problem.csv')
Y = read_column_from_csv(1 + (v_number - 1) * 2, 'data/7problem.csv')
draw_plot(X, Y)
for k in range(2, 5):
clusters = clustering(k, X, Y)
draw_clusters(clusters)
import pandas as pd
import statsmodels.formula.api as smf
import math
from lib import rss, ess
from scipy.stats import f, norm, chi
import numpy as np
from lib import mk_data_var, read_column_from_csv
# TODO alter this to your variant
v_number = 1
mk_data_var(v_number)
class_1 = read_column_from_csv(0,
'data/6problem_{}.csv'.format(v_number),
type='f')
class_2 = read_column_from_csv(1,
'data/6problem_{}.csv'.format(v_number),
type='f')
sex = read_column_from_csv(3,
'data/6problem_{}.csv'.format(v_number),
type='f')
survived = read_column_from_csv(4,
'data/6problem_{}.csv'.format(v_number),
type='f')
df = pd.DataFrame({
"class_1": class_1,
"class_2": class_2,
"sex": sex,
d3=1, если квартира трёхкомнатная, 0 иначе;
d4=1, если квартира четырёхкомнатная, 0 иначе;
dist расстояние от центра Москвы (в км);
walk=1, если до метро можно быстро дойти пешком, 0 иначе;
brick=1, если дом кирпичный, 0 иначе;
bal=1, если есть балкон, 0 иначе;
floor=0, если этаж первый или последний, 1 иначе.
"""
variation = 1
# TODO Вы должны сделать свой .csv файл из того что прислал Фурманов. Нужно удалить из него все строки
# в которых есть пустые элементы. Пустые строки в каждом варианте разные, поэтому удалите
# только те, что пустые именно в вашем варианте. После этих операций сохраните его как data/5problem.csv
bal = read_column_from_csv(column_number=0 + (variation - 1) * 11,
file='data/5problem.csv')
brick = read_column_from_csv(column_number=1 + (variation - 1) * 11,
file='data/5problem.csv')
d2 = read_column_from_csv(column_number=2 + (variation - 1) * 11,
file='data/5problem.csv')
d3 = read_column_from_csv(column_number=3 + (variation - 1) * 11,
file='data/5problem.csv')
d4 = read_column_from_csv(column_number=4 + (variation - 1) * 11,
file='data/5problem.csv')
dist = read_column_from_csv(column_number=5 + (variation - 1) * 11,
file='data/5problem.csv')
floor = read_column_from_csv(column_number=6 + (variation - 1) * 11,
file='data/5problem.csv')
price = read_column_from_csv(column_number=7 + (variation - 1) * 11,
file='data/5problem.csv')
totsp = read_column_from_csv(column_number=8 + (variation - 1) * 11,
методом наименьших квадратов""")
print()
# LS(least squares)
# b1 b2 b3 b4
coefficient_vector = ls(variation)
print("Коэфициенты b1 b2 b3 b4:")
print(coefficient_vector)
print()
##############################################################
######## Проверьте значимость регрессии в целом ##############
##############################################################
print("2. Проверьте значимость регрессии в целом")
print()
x_2 = read_column_from_csv(column_number=0 + (variation - 1) * 4,
file='data/4problem.csv')
x_3 = read_column_from_csv(column_number=1 + (variation - 1) * 4,
file='data/4problem.csv')
x_4 = read_column_from_csv(column_number=2 + (variation - 1) * 4,
file='data/4problem.csv')
y_1 = read_column_from_csv(column_number=3 + (variation - 1) * 4,
file='data/4problem.csv')
y_estimation = [
coefficient_vector[0] + coefficient_vector[1] * x_2[i] +
coefficient_vector[2] * x_3[i] + coefficient_vector[3] * x_4[i]
for i in range(40)
]
ess_ur = ess(y_1, y_estimation)
rss_ur = rss(y_1, y_estimation)
методом наименьших квадратов""")
print()
# LS(least squares)
# b1 b2 b3 b4
coefficient_vector = ls(variation)
print("Коэфициенты b1 b2 b3 b4:")
print(coefficient_vector)
print()
##############################################################
######## Проверьте значимость регрессии в целом ##############
##############################################################
print("2. Проверьте значимость регрессии в целом")
print()
x_2 = read_column_from_csv(column_number=0 + (variation - 1) * 4,
file=file_path)
x_3 = read_column_from_csv(column_number=1 + (variation - 1) * 4,
file=file_path)
x_4 = read_column_from_csv(column_number=2 + (variation - 1) * 4,
file=file_path)
y_1 = read_column_from_csv(column_number=3 + (variation - 1) * 4,
file=file_path)
y_estimation = [
coefficient_vector[0] + coefficient_vector[1] * x_2[i] +
coefficient_vector[2] * x_3[i] + coefficient_vector[3] * x_4[i]
for i in range(40)
]
ess_ur = ess(y_1, y_estimation)
rss_ur = rss(y_1, y_estimation)