Page MenuHomec4science
Files F120821468

helpers_plot.py
No OneTemporary
Actions

File Metadata

Created
Mon, Jul 7, 08:19

helpers_plot.py
View Options

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from datetime import timedelta
def format_energy_plot(y_min, y_max):
"""
Utility to format all time series plots
:param double y_min: minimum y values displayed
:param double y_max: maximum y values displayed
"""
plt.gca().set_ylim([y_min, 1.2*y_max]) # TODO: fix y_max to add also legend
plt.xticks(rotation=20)
plt.gca().set_ylabel('Energy consumption [kWh]')
plt.grid('True')
def plot_clusters(clustering, ordered_clusters, PCA_weeks, outliers, building_number=None):
"""
Plot clusters,every point is a week: normal weeks are in blue, outliers are in black
:param output of DBSCAN clustering: results of the clustering made using DBSCAN function
:param matrix ordered_clusters: dimension number of cluster x number of weeks, collocate evry week in a cluster
:param matrix PCA_weeks: dimension principal_components_number x total_weeks, contains per every week the first indicated principal components
:param vector outliers: length numebr of weeks, indicate if a week is an outlier
:param int building_number: number of the considered building
"""
max_cluster_index = np.max(clustering.labels_)
# Scatterplot the result
for cluster_index in range(0, max_cluster_index+1):
plt.scatter(PCA_weeks[0, ordered_clusters[cluster_index]], PCA_weeks[1, ordered_clusters[cluster_index]])
plt.scatter(PCA_weeks[0, outliers], PCA_weeks[1, outliers], color='black')
if building_number != None:
plt.title('Building ' + str(building_number))
plt.gca().set_xlabel('PC1')
plt.gca().set_ylabel('PC2')
plt.show()
def plot_clustered_weeks(total_weeks, data_per_week, ordered_clusters, energy_consumptions):
"""
Plot time serie in groups of 4 weeks using different colors to distinghuish between the clustered weeks:
blue -> normal weeks
yellow, black -> atypical weeks
:param int total_weeks: numebr of weeks in the dataset (default value is 98)
:param data_per_week: number of measurement for every week (default value is 672, a measurement every 15 minutes)
:param matrix ordered_clusters: dimension number of cluster x number of weeks, collocate evry week in a cluster
:param DataFrame energy_consumptions: formatted DataFrame containing measurements of energy consumption.
Index is of type DatetimeIndex,
column 'consumption' contains the measurement,
column 'week' contains the week code of the measurement.
"""
data_per_group = data_per_week*4
group_range = [0, int(total_weeks/4)]
week_indices = np.arange(0, total_weeks)
for group_index in range(group_range[0], group_range[1]):
plt.figure(figsize=(20, 4))
for local_week_index in range(0, 4):
global_week_index = group_index*4 + local_week_index
# Look for the current week in every cluster (plot with relative color)
found_in_cluster = False
for cluster_index in range(0, len(ordered_clusters)):
if (global_week_index in week_indices[ordered_clusters[cluster_index]]):
week_color = 'C' + str(cluster_index)
energy_consumptions.loc[energy_consumptions['week'] == global_week_index, 'consumption']\
.plot(grid='True',rot=45, color=week_color)
found_in_cluster = True
# If week not found then it's outlier (plot in black)
if not found_in_cluster:
energy_consumptions.loc[energy_consumptions['week'] == global_week_index, 'consumption']\
.plot(grid='True', rot=45, color='black')
plt.title('Group ' + str(group_index))
format_energy_plot(energy_consumptions.consumption.min(),
energy_consumptions.consumption.max())
plt.show()
def plot_model_week(model_week, data_per_week, energy_consumptions):
"""
Plot the model of a "standard" week, computed using the median value for every measurement in th week
:param vector model_week: model
:param data_per_week: number of measurement for every week
:param DataFrame energy_consumptions: formatted DataFrame containing measurements of energy consumption.
Index is of type DatetimeIndex,
column 'consumption' contains the measurement,
column 'week' contains the week code of the measurement.
"""
points = np.arange(data_per_week)
plt.plot(points, model_week, color='C1')
format_energy_plot(energy_consumptions.consumption.min(),
energy_consumptions.consumption.max())
plt.locator_params(axis='x', nbins=7)
plt.title("Model week")
plt.show()
def plot_regularized_weeks(regularized_energy_consumptions, atypical_weeks, total_weeks, data_per_week = 672):
"""
Plot the weeks that have been regularized, i.e. the days with a large number of anomalies have been substituted with the corresponging day of themodel week
:param DataFrame regularized_energy_consumptions: regularized DataFrame containing measurements of energy consumption.
Index is of type DatetimeIndex,
column 'consumption' contains the measurement,
column 'week' contains the week code of the measurement.
:param vector atypical_weeks: indexes of atypical weeks
:param int total_weeks: numebr of weeks in the dataset
:param data_per_week: number of measurement for every week (default value is 672, a measurement every 15 minutes)
"""
rows = np.arange(data_per_week)
for i, global_week in enumerate(atypical_weeks):
regularized_energy_consumptions.iloc[rows+global_week*data_per_week, 0]\
.plot(grid='True',rot=45)
plt.title('Week ' + str(atypical_weeks[i]))
format_energy_plot(regularized_energy_consumptions.consumption.min(),
regularized_energy_consumptions.consumption.max())
plt.show()
def plot_coefficients_lr(model, X_train):
"""
Plots sorted coefficient values of the model
:param model: linear regression model
:param matrix X_train: data used for training of the linear regression model
"""
coefs = pd.DataFrame(model.coef_, X_train.columns)
coefs.columns = ["coef"]
coefs["abs"] = coefs.coef.apply(np.abs)
coefs = coefs.sort_values(by="abs", ascending=False).drop(["abs"], axis=1)
plt.figure(figsize=(15, 7))
coefs.coef.plot(kind='bar')
plt.grid(True, axis='y')
plt.hlines(y=0, xmin=0, xmax=len(coefs), linestyles='dashed')
def plot_anomalies(regression_dataframe, groups_number = 24,save_fig = False):
"""
Plot time series in group of 4 weeks, real values are in blue, model is orange and red points are the detected anomalies
:param DataFrame regression_dataframe: DataFrame containing measurements of energy consumption.
Index is of type DatetimeIndex,
column 'consumption' contains the real measurement
column 'model' contains the linear regression model
column 'anomalies' contains a bool indicating if the point is anomalous
"""
# Plot data in groups of 4 weeks
# number of weekly measurements: 672
# number of full available weeks: total_weeks
data_per_group = 672*4
group_range = [0, groups_number]
rows = np.arange(data_per_group)
for i in range(group_range[0], group_range[1]):
plt.figure(figsize=(20, 4))
ts_for_plot = regression_dataframe.iloc[rows+i*data_per_group, 0]
plt.plot(pd.to_datetime(ts_for_plot.index), ts_for_plot.values)
ts_for_plot = regression_dataframe.iloc[rows+i*data_per_group, 1]
plt.plot(pd.to_datetime(ts_for_plot.index), ts_for_plot.values)
anomalies_bool = regression_dataframe.iloc[rows+i*data_per_group, 2].values
anomalies_time = regression_dataframe.index[i*data_per_group + np.where(anomalies_bool == True)[0]]
anomalies_value = regression_dataframe.iloc[i*data_per_group +np.where(anomalies_bool == True)[0], 0].values
plt.plot(pd.to_datetime(anomalies_time), anomalies_value, 'or')
plt.title('Group ' + str(i))
format_energy_plot(regression_dataframe.consumption.min(),
regression_dataframe.consumption.max())
if (save_fig):
plt.savefig('Anomalies' + str(i) +'.png')
plt.show()
def plot_weeks(results_linear_regression, weeks_number, first_week_index, figure_size=(20, 4), image_name='', data_per_week=672):
"""
Plot time series of a weeks_number weeks, real values are in blue, model is orange and red points are the detected anomalies
:param DataFrame regression_dataframe: DataFrame containing measurements of energy consumption.
Index is of type DatetimeIndex,
column 'consumption' contains the real measurement
column 'model' contains the linear regression model
column 'anomalies' contains a bool indicating if the point is anomalous
:param int weeks_number: how many weeks to plot
:param list first_week_index: index of the first weeks to plot (warning: no check on last weeks!)
:param tuple figure_size: speccifies the size of the plot
:param string image_name: name of the file in which the image is saved (image not saved if no name provided)
:param data_per_week: number of measurement for every week (default value is 672, a measurement every 15 minutes)
"""
# Set figure size
plt.figure(figsize=figure_size)
# Build the vector containing measurements indices
rows = np.arange(data_per_week)
time_indices = rows + first_week_index*data_per_week
for week_index in range(first_week_index + 1, first_week_index + weeks_number):
time_indices = np.hstack((time_indices, rows + week_index*data_per_week))
# Plot both real values and model
for column_index in range(2):
ts_for_plot = results_linear_regression.iloc[time_indices, column_index]
plt.plot(pd.to_datetime(ts_for_plot.index), ts_for_plot.values)
# Plot anomalies
anomalies_bool = results_linear_regression.iloc[time_indices, 2].values
anomalies_time = results_linear_regression.index[first_week_index*data_per_week + np.where(anomalies_bool == True)[0]]
anomalies_value = results_linear_regression.iloc[first_week_index*data_per_week +np.where(anomalies_bool == True)[0], 0].values
plt.plot(pd.to_datetime(anomalies_time), anomalies_value, 'or')
format_energy_plot(results_linear_regression.consumption.min(),
results_linear_regression.consumption.max())
plt.legend(['Measurements', 'Model', 'Anomalies'], loc='upper right')
if image_name != '':
plt.savefig(image_name + '.png')
plt.show()

Event Timeline

c4science · Help