Week 7 – Post Mortem

--Originally published at TC3045 – Sagnelli's blog

Howdy partner,

As you may have read, or maybe not, in the pre mortem blog for this week, I worked on being able to mine tweets from an account since the beginning of their time. I was able to do that by creating a loop in which 200 tweets were gathered using the API of Twitter. Also it is possible to specify the people that you want to gather tweets from by writing their user names in a txt file stored in the same location.

This time we want to gather all the tweets from @RicardoAnayaC , @lopezobrador_ , and @JoseAMeadeK to create a words map. This map will allow us to know which words are most used by the candidates to measure their popularities.

Stay tuned, for the next weeks’ progress


Election year: let’s be genuine, shall we?

--Originally published at TC3045 – Sagnelli's blog

2018 is an important year for Mexico, where the next six years are supposed to be defined by the Mexican people; however, corruption has always interfered with democracy, as the government has been accused of manipulating the votes.

Resultado de imagen para simpsons vote gif

This is the problem we are trying to solve with our project. Now, the important question is:

Who are we?

We are a group of students between 6th and 8th semester of Computer Science at Tec de Monterrey Campus Guadalajara:

  • Alfonso Contreras
  • Arturo González
  • Alejandro Vázquez
  • Michelle Sagnelli

What is our solution?

Basically, in one sentence, we are building a series of microservices that will let us determine who is the best acclaimed, and the most popular presidential candidate according to Twitter.

How are we supposed to do it?

We will apply data mining using Python Streaming Jobs, and Twitter’s API to temporarily store tweets in JSON’s. Afterwards, this data will be shown and saved for later use.

The challenge is to clean data by mining keywords, eliminating stopwords, and assigning tokens by tweet importance. Henceforth, this “clean” data will be used to analyze with machine learning the importance of this year’s candidates, and political parties. Finally, this information will be stored in JSON format for further analysis of political parties information, and candidates’ level of acceptance.


We are trying to implement location-based analysis of tweets, and being able to find which tweets belong to bots for achieving a more successful analysis.

This should be fun. I am very interested in this project, as it is challenging, and interesting. If you are interested too, do not hesitate in contacting me, and stay tuned with mine, and my colleagues’ , future blog posts.