jiegzhan / time-series-forecasting-rnn-tensorflow

Time series forecasting

Hi there 👋

• 🔭 I am Zhang Jie (张 杰).

• In 05/2024, I joined Disney Streaming as a Staff Software Engineer, where I mainly focus on data ingestion, data discovery, and data governance.

• I was a Senior Software Engineer at Roku Big Data Platform team for 4.5 years, where I provided data infrastructure and data solutions both in large scale real time streaming processing and data lake batch processing.

• Tech Stack: Flink, Spark, Kafka, Presto, Iceberg, Hive, Hadoop, Airflow, Kubernetes, Docker, AWS Stack, DataDog, Jupyter Notebook, Superset, Looker.

Real Time Streaming Processing ✅

Built a Flink & Kubenetes powered real time streaming platform which provides capabilities to build Flink streaming applications and run them on Kubernetes clusters seamlessly. Onboarded other engineering teams and promoted best streaming practices.

Data Lake Batch Processing ✅

Built and maintained a Spark & Hive & S3 & Airflow based data lake, architected and implemented distributed data ingestion and processing pipelines.

time-series-forecasting-rnn-tensorflow's Issues

"Predicted" and "Actual" labels on the final chart are inversed

In train_predict.py there is a mix up with labels. What should be "Actual" is shown as "Predicted" and vice versa:

        # Plot graph: predicted VS actual
        plt.subplot(111)
        plt.plot(predicted_raw, label='Actual')
        plt.plot(y_test_raw, label='Predicted')
        plt.legend()
        plt.show()

what is the parameter_file ?

what dose that file include?

Why did you use model.fit for training?

To train a model in Tensorflow I checked other sources they use sessions in loop in order to train model.

with tf.Session() as sess:
   init.run()
   for ep in range(epochs):
         sess.run(training_op, feed_dict={X: x_batches, y:y_batches})
         mse = loss.eval(feed_dict={X: x_batches, y:y_batches})

Why did you use model.fit?

Multiple inputs

Hi,

I want to use your model for a time-series that has 6 inputs and just 1 output, can you help me? I think i have to upload the file with 6 columns, but i dont know how to read it and manage all this inputs.

Thank you.

Why you shuffle the sequence of training data?

In the pre- processing before entering network model, why you shuffle the sequence of training data?
Since the sequence is important in time series!

predict for next 5 timesteps

what if I want to predict for next 5 timesteps ?