README.md

Making Predictions with Temperature Data

• Setting up Python
• Exploring Time Series Data
• Making Univariate Predictions with Temperature Data
  • Predictions with SARIMA
  • Predictions with Basic Neural Network
  • Predictions with LSTM
• Making Multivariate Predictions

Setting up Python

To get started you will need git installed on your computer.

https://git-scm.com/book/en/v2/Getting-Started-Installing-Git

Change to your local user directory.

Let's put our code in a local user directory named 'repo' (You don't have to use the name repo, or any name you like) . Create the directory by using the following commands:

$ cd
mkdir repo

Change to the 'repo' directory that you just created. This is where you will place your code.

cd repo

Inside the code directory, clone the repository for this workshop:

repo$ git clone https://github.com/pyladieshamburg/getting-started-raspberry-pi.git

Change directory, so that you are inside the cloned repository:

repo$ cd getting-started-raspberry-pi 

Create a python virtual environment. The name of the virtual environment is 'env'. You may name the environment whatever you like.

getting-started-raspberry-pi$ python3 -m venv env

Activate the newly created virtual environment

getting-started-raspberry-pi$ source env/bin/activate

You have successfully activated your environment when the name of your environment is prefixed at the command line:

(env) getting-started-raspberry-pi$

Now we need to install the required packages to use the code.

(env) getting-started-raspberry-pi$  pip install -r requirements.txt

We must tell Jupyter notebook how to find your python kernel:

(env) getting-started-raspberry-pi$  pip install tornado
(env) getting-started-raspberry-pi$  pip install ipykernel
# Create the kernel (from within the virtual environment that we created above)
(env)$ python -m ipykernel install --user --name env --display-name "PyLadies-Time-Series"

Download data we will use for the workshop:

https://secure.sonnenstein.org/nextcloud/index.php/s/GPB4fg4xcFMcv58

https://secure.sonnenstein.org/nextcloud/index.php/s/Zmp4GnRh7N8S3ov

Let's look at some code. Start your Jupyter notebook:

(env) getting-started-raspberry-pi$  jupyter-lab

Exploring Temperature and Humidity Data

In the data-exploration exercise, you create and visualize time series data.

You also examine some important characteristics of time series such as seasonality and trend ad conclude by about learning commons tools for time series analysis.

Making Univariate Predictions with Temperature Data

In these exercises you will experiment with different approaches to making univariate time series with predictions.

Before you build use complex techniques for univariate time series prediction, traditional approaches should also be implemented as a baseline. The predictions-with-sarima you focus on building a SARIMA Model and different ways to select SARIMA hyperparameters.

In the predictions-with-neural-network exercise, you will set up a basic LSTM and experiment with options for training the best model.

In the predictions-with-lstm exercise, you will set up a basic neural network and experiment with options for training the best model.

Making Multivariate Predictions

The prediction-season exercise - for fun - you use both temperature and humidy to predict the season of the year.