Skip to content

Commit

Permalink
refactor: refactored plots
Browse files Browse the repository at this point in the history
  • Loading branch information
@roscibely
roscibely committed Sep 26, 2023
1 parent c289838 commit 007f358
Show file tree
Hide file tree
Showing 2 changed files with 85 additions and 27 deletions.
52 changes: 41 additions & 11 deletions period_cycle_prediction/models/linear_regression.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@

if __name__ == '__main__':
# Abrir dataset sintético
df = pd.read_csv('period_cycle_prediction\dataset\synthetic_data.csv', sep=',', header=0)
df = pd.read_csv('period_cycle_prediction/dataset/synthetic_data.csv', sep=',', header=0)
# Preparando os dados para o modelo de regressão linear/machine learning
periods_data = utils.calculate_datatime(df)
features, labels = utils.generate_final_features(df)
Expand Down Expand Up @@ -46,28 +46,46 @@
# predição um passo a frente / novo ciclo
predicao_um_passo_a_frente = model_LR.predict([test_x[-1]])
cycles_numbers = np.arange(1, len(cycle_length)+1)





plt.figure(figsize=(4,4))
width = 8
height = 5
plt.figure(figsize=(width, height))
plt.rcParams.update({'font.size': 16})

plt.plot(cycle_length, '-->', color='blue')
plt.plot(periods, '-*', color='green')
plt.plot(test_y, ':o', color='red')
plt.plot(cycle_length, '-->', color='blue', linewidth=3.0)
plt.plot(test_y[:,0], ':o', color='red', linewidth=3.0 )
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][0], '-->', color='blue')
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][1], '-*', color='green')
plt.legend(['Duração dos ciclo (Predito)', 'Variação dos periodos (Predito)', 'Dados reais'])
plt.legend(['Predito', 'Real', 'Predição um passo a frente'])
plt.grid()
plt.xlabel('Ciclos')
plt.ylabel('Dias')
plt.title('Modelo Regressão Linear')
plt.title('Variação dos Ciclos')
fig = plt.gcf()
fig.savefig('linear.png', dpi=300, bbox_inches='tight')
plt.show()

error = abs(test_y-y_pred)
plt.figure(figsize=(width, height))
plt.plot(periods, '-*', color='green', linewidth=3.0)
plt.plot(test_y[:,1], ':o', color='red', linewidth=3.0 )
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][1], '-*', color='green')
plt.legend(['Predito', 'Real', 'Predição um passo a frente'])
plt.grid()
plt.xlabel('Períodos')
plt.ylabel('Dias')
plt.title('Variação dos Períodos')
fig = plt.gcf()
fig.savefig('linear_periods.png', dpi=300, bbox_inches='tight')
plt.show()

plt.figure(figsize=(width, height))
output_pred = [[int(round(i[0])), int(round(i[1]))] for i in y_pred] # round the values
error = abs(test_y-output_pred)
plt.plot(error[:,0], '-->', color='blue')
plt.plot(error[:,1], '-*', color='green')
plt.legend(['Erro Ciclo', 'Erro Período'])
plt.legend(['Erro Ciclo'])
plt.grid()
plt.xlabel('Ciclos')
plt.ylabel('Dias')
Expand All @@ -76,6 +94,18 @@
fig.savefig('linear_error.png', dpi=300, bbox_inches='tight')
plt.show()

plt.figure(figsize=(width, height))

plt.plot(error[:,1], '-*', color='green')
plt.legend(['Erro Período'])
plt.grid()
plt.xlabel('Períodos')
plt.ylabel('Dias')
plt.title('Modelo Regressão Linear')
fig = plt.gcf()
fig.savefig('linear_error_periods.png', dpi=300, bbox_inches='tight')
plt.show()



# calcular o RMSE
Expand Down
60 changes: 44 additions & 16 deletions period_cycle_prediction/models/random_forest.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@

if __name__ == '__main__':
# Abrir dataset sintético
df = pd.read_csv('period_cycle_prediction\dataset\synthetic_data.csv', sep=',', header=0)
df = pd.read_csv('period_cycle_prediction/dataset/synthetic_data.csv', sep=',', header=0)
# Preparando os dados para o modelo
periods_data = utils.calculate_datatime(df)
features, labels = utils.generate_final_features(df)
Expand Down Expand Up @@ -45,39 +45,67 @@
predicao_um_passo_a_frente = model_RD.predict([test_x[-1]])
cycles_numbers = np.arange(1, len(cycle_length)+1)

plt.figure(figsize=(4,4))
width = 8
height = 5
plt.figure(figsize=(width, height))
plt.rcParams.update({'font.size': 16})

plt.plot(cycle_length, '-->', color='blue')
plt.plot(periods, '-*', color='green')
plt.plot(test_y, ':o', color='red')

plt.plot(cycle_length, '-->', color='blue', linewidth=3.0)
plt.plot(test_y[:,0], ':o', color='red', linewidth=3.0 )
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][0], '-->', color='blue')
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][1], '-*', color='green')
plt.legend(['Duração dos ciclo (Predito)', 'Variação dos periodos (Predito)', 'Dados reais'])
plt.legend(['Predito', 'Real', 'Predição um passo a frente'])
plt.grid()
plt.xlabel('Ciclos')
plt.ylabel('Dias')
plt.title('Modelo Floresta Aleatória')
plt.title('Variação dos Ciclos')
fig = plt.gcf()
fig.savefig('random.png', dpi=300, bbox_inches='tight')
fig.savefig('rf.png', dpi=300, bbox_inches='tight')
plt.show()

error = abs(test_y-y_pred)
plt.figure(figsize=(width, height))
plt.plot(periods, '-*', color='green', linewidth=3.0)
plt.plot(test_y[:,1], ':o', color='red', linewidth=3.0 )
plt.plot(cycles_numbers[-1], predicao_um_passo_a_frente[0][1], '-*', color='green')
plt.legend(['Predito', 'Real', 'Predição um passo a frente'])
plt.grid()
plt.xlabel('Períodos')
plt.ylabel('Dias')
plt.title('Variação dos Períodos')
fig = plt.gcf()
fig.savefig('rf_periods.png', dpi=300, bbox_inches='tight')
plt.show()

plt.figure(figsize=(width, height))
output_pred = [[int(round(i[0])), int(round(i[1]))] for i in y_pred] # round the values
error = abs(test_y-output_pred)
plt.plot(error[:,0], '-->', color='blue')
plt.plot(error[:,1], '-*', color='green')
plt.legend(['Erro Ciclo', 'Erro Período'])
plt.legend(['Erro Ciclo'])
plt.grid()
plt.xlabel('Ciclos')
plt.ylabel('Dias')
plt.title('Modelo Floresta Aleatória')
plt.title('Modelo Random Forest')
fig = plt.gcf()
fig.savefig('random_error.png', dpi=300, bbox_inches='tight')
fig.savefig('rf_error.png', dpi=300, bbox_inches='tight')
plt.show()

plt.figure(figsize=(width, height))

plt.plot(error[:,1], '-*', color='green')
plt.legend(['Erro Período'])
plt.grid()
plt.xlabel('Períodos')
plt.ylabel('Dias')
plt.title('Modelo Random Forest')
fig = plt.gcf()
fig.savefig('rf_error_periods.png', dpi=300, bbox_inches='tight')
plt.show()



# calcular o RMSE
rms = sqrt(mean_squared_error(test_y, y_pred))
print('RMSE: ', rms)

# calcular o MAE (Mean Absolute Error)
mae = np.mean(np.abs((test_y - y_pred)))
print('MAE: ', mae)
print('MAE: ', mae)

0 comments on commit 007f358

Please sign in to comment.