felipecruz91 / twitter-analytics-pi

Twitter Analytics running in a Raspberry Pi 3 Model B

twitter-analytics-pi

Twitter Analytics running in a Raspberry Pi 3 Model B

Overview

.NET Core 2.2 Console App which listens for tweets given a keyword on the Twitter Stream feed, performs a sentiment analysis of the tweet using Azure Text Analytics API and stores the tweet and the sentiment score in an Influx DB database.

Prerequisites

Hardware

• A Raspberry Pi 3 Model B with Raspbian already installed.

Software

• A Docker installation in your Raspberry Pi. Tutorial
• An Influx DB server running in your Raspberry Pi as a Docker container.

Running your InfluxDB image

Start your image binding the external port 8086 of your containers:

docker run -d -p 8086:8086 hypriot/rpi-influxdb

Twitter Analytics configuration setup

Set up the following configuration in the appsettings.json and appsettings.*.json files with your own values:

{
  "InfluxDB": {
    "ServerAddress": "http://<INFLUX_DB_SERVER_ADDR>:<INFLUX_DB_PORT>"
  },
  "Twitter": {
    "ConsumerKey": "",
    "ConsumerSecret": "",
    "AccessToken": "",
    "AccessTokenSecret": ""
  },
  "TextAnalytics": {
    "Name": "",
    "Key1": "",
    "Key2": ""
  } 
}

Build the Twitter Analytics Docker image

bash build.sh

Upload the Docker image to your public or private registry

bash push.sh

Running the container from Twitter Analytics Docker image in the Raspberry Pi

Log into your Raspberry Pi by ssh:

λ ssh pi@<YOUR_RASPBERRY_IP_ADDR>

Run the container

docker run --rm -e "keyword=climate change" youraccount/twitteranalytics

The container will start listening for tweets that contain the word climate change on the Twitter Stream feed and will store them in an Influx DB database.

See the tweets and sentiment score in your Influx DB

Go inside your Influx DB container:

λ docker exec -it <INFLUX_DB_CONTAINER_ID> influx

> use TwitterAnalytics

• The tweet information is stored in the tweet measurement.

> SELECT * FROM tweet

• The sentiment information is stored in the sentiment measurement.

You can see the average sentiment score of your tweets running the following query:

> SELECT MEAN(score) FROM sentiment

name: sentiment
time mean
---- ----
0    0.47096266531824166