search

sherlockhoatszx / TimeSeriesPredctionUsingDeeplearning

Try to make Time Series Prediction with simple neurons network supported by Keras
4 stars 5 forks source link

Question: The accuracy is very low #1

Open KP-Zhang opened 7 years ago

KP-Zhang commented 7 years ago

Hi, Sir, Your work is very impressive. I am quite new to keras. When I run your code, I add a parameter -- metrics=['accuracy'] in the compile function. But the acc is unreasonably low. I do not know why. What happened? Can you help me out of this confusion? Thank you for your time.

sherlockhoatszx commented 7 years ago

hi Dylan: My accuracy metrics (predict the next 2 days) will be around 85-93%. I have upgrade my code by adding one Convolution1D layers before feed into the GRUorLSTM layers. You can try to use Convolution1D before it, and fine tune your hyper parameter .Hope that can help.

在 2017年2月8日，上午4:29，DylanCheung notifications@github.com 写道：

Hi, Sir, Your work is very impressive. I am quite new to keras. When I run your code, I add a parameter -- metrics=['accuracy'] in the compile function. But the acc is unreasonably low. I do not know why. What happened? Can you help me out of this confusion? Thank you for your time.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sherlockhoatszx/TimeSeriesPredctionUsingDeeplearning/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/AOM_eC6i2gmAc0rTTzvnlsa2hxyOqs-3ks5raNREgaJpZM4L5_ok.

KP-Zhang commented 7 years ago

Hi sherlockhoatszx, Thank you for your reply. Forgive me stupid question. I am not quite sure whether I apply a right way to calculate the 'accuracy'. If I simply modify your code: model.compile(loss='mean_squared_error', optimizer=Adam()) TO model.compile(loss='mean_squared_error', optimizer=Adam(), metrics=['accuracy']), does it work? How do you calculate your accuracy metrics? Happy Lantern Festival!

KP-Zhang commented 7 years ago

Hi Sherlockhoatszx, Do you have a moment? model.compile(loss='mean_squared_error', optimizer=Adam()) TO model.compile(loss='mean_squared_error', optimizer=Adam(), metrics=['accuracy']), does it work? How do you calculate your accuracy metrics?

sherlockhoatszx commented 7 years ago

I use MAPE to calculate it . MAPE function was wrote by myself.

KP-Zhang commented 7 years ago

Thank you for your reminding. I am trying to do MAPE based on this site http://stats.stackexchange.com/questions/58391/mean-absolute-percentage-error-mape-in-scikit-learn/62511#62511. Thank you for your time.

KP-Zhang commented 7 years ago

Hi Sherlockhoatszx, it is always helpful to read your code. I have some confusion related to Time Series Forecasting. There is traffic data (1440 pieces in total, and 288 pieces each day) I collected to predict traffic flow. The data is collected every 5 min in five consecutive working days. I am going to use the traffic data of the first four day to train the prediction model, while the traffic data of the fifth day is used to test the model. Here is my question, if I want to predict the traffic flow of the fifth day, do I only need to treat my prediction as one-step forecast or do I have to predict 288-step? Look forward to your advice. Thanks for your post again.