I recently made a presentation to aspiring data scientists about one of my first project for a client: “money laundering detection”.
Today, I share this story with you, so you can learn more about data consultant job:
Today, I share this story with you, so you can learn more about data consultant job:
That was 8 years ago, I was working in a small team (<10 persons) with various profiles (data engineer, data scientist, business consultant…), and we followed CRIP-DM methodology in agile mode.
(I explained CRISP-DM earlier here:)
(I explained CRISP-DM earlier here:)
What does it mean?
I listened to my client needs and I asked him a lot of questions regarding his business, his customers, and data available. This period took a few weeks.
I listened to my client needs and I asked him a lot of questions regarding his business, his customers, and data available. This period took a few weeks.
Once I got data I needed, I explored them. Made statistics. I used IBM SPSS Modeler for this.
I wasn’t good at python at that time, and this software is amazing for non-coders. You can easily explore data and create algo with it.
I wasn’t good at python at that time, and this software is amazing for non-coders. You can easily explore data and create algo with it.
We established with the client a persona of someone who do laundry money and we decided on algorithms to detect these personas.
While we built algorithms, at regular interval we planned meetings with client to show him intermediate results and ask questions to be sure we were building what he expects (agile methodology).
”Showing results” is often a powerpoint with graph, stats, facts we discovered while digging into data, questions. This period lasted a few months.
Finally, when the algo was doing great and the client validated it, we entered in a deployment phase. To illustrate a deployment: we built a sand castle and now we wanted to build a real castle.
For example, your code work on your computer but you want people to use it over internet. You need to deploy& integrate it.
This phase was longer. I worked on the transition, then data engineer/developers continued the deployment and I left for another nice project.
This phase was longer. I worked on the transition, then data engineer/developers continued the deployment and I left for another nice project.
That’s consultant life, you work for a client, and then you leave for another one. I like it because you can see a lot of different projects, techno business. And sometimes it’s frustrating because you can leave a nice team.
Well, that's it, I hope you now have an idea of what a consultant in data science can do (ask questions, build, show results...).
1. Follow me @Pauline_Cx to learn more about data & SaaS
2. RT the tweet below to share my story with your audience
1. Follow me @Pauline_Cx to learn more about data & SaaS
2. RT the tweet below to share my story with your audience
Loading suggestions...