This sample uses functions to forecast temperatures based on a series of temperature data. It uses statsmodel autoregression to retrain the data.

• Install Python 3.6+
• Install Functions Core Tools
• Install Docker
• Note: If run on Windows, use Ubuntu WSL to run deploy script

• Click Deploy to Azure Button to deploy resources

or

• Deploy through Azure CLI

  • Open AZ CLI and run az group create -l [region] -n [resourceGroupName] to create a resource group in your Azure subscription (i.e. [region] could be westus2, eastus, etc.)
  • Run az group deployment create --name [deploymentName] --resource-group [resourceGroupName] --template-file azuredeploy.json

• Download dataset from here https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv

• Deploy Function App

  • Create/Activate virtual environment
  • Run func azure functionapp publish [functionAppName] --build-native-deps

• Upload the csv dataset to the forecastinput container blob either through portal or through following Azure CLI

az storage blob upload --container-name forecastinput --account-name {storageName} -f {dataset} -n daily-minimum-temperatures.csv

• Send the following body in a HTTP POST request as a query param where name: Input CSV file result: Forecast output graph image

http://[functionappname]/api/ForecastAPI?name=daily-minimum-temperatures.csv&result=series.png

For any local testing, use the sample local.settings.json and host.json, create virtual environment and run func host start

• Create your first Python Function
• Time Series Autoregression Model