Cool title: Ideas to set up a data science team in 8 weeks.

Notes: The importance of powerpoints is often underestimated by the other data scients. I found myself spending almost half the time in polishing results rather than working on new ones.

Time sensitivity. I felt I could have managed better the effort in the start of the project. This is a two way street, also mamacrowd should have prepared the database description in advance. Not having the database description ready factor for a lot of disadvantage as it broke momentum from excitment of the start.

Leveraging start excitmenet is probablyy very importnat.

A lo t of tools were presented by the project Drive + Asana + Various formats for slides but i feel that the various productivity tool were not need. Working in a small team allows for fast direct communication to be the best way to move forward This does not mean that goals were not USEFUL. It is important in the direct communicatoin to know how to align each other on the work done.

PROBABLY TH EBIGGEST OSTACLE IS TO FIND THE TIME TO REVIEW WHAT YOUR OTHER TEAMMATE IS DOING. In such fastpaced project one is very often involved in its own analysis and even if there is just a minor difference from the other the development deploys in two different directoins. This is a huge waste of time.

I found myself confused after 3 weeks that my teammated made a similar analysis in week 1!!!. I found myself also not involved in looking at other members work if not before presentation to stakeholder, which i think was a mistake. I found myself wasting too much time in the beginning find a way to navigate the db and in retrospective does not seem valid The best experience was conducting the analysis and step by step tracking metrics such as number of users total amount spent in different stage of the analysis process, and this gave a me an intuition. When I merged with other features such as campaign names it gave me insight( regarding the effect of marketing with campaigns such as lifegate and forno brisa, i had an headsup from the company regarding this being highly influential factors but the data spoke by itself.)

I started with checking my results with the outputted from the company, but i eventually forgot as the analysis got more involved, and again in week 4 i found myself finding that i was actually evaluating only half of the total amount!!

I was praised in being very clear in exposition :D

I found very interesting the observatoin of being smart in what to show, you do not want to show too much too fast and the spend the rest of the time trying to calibrate that or it gives the impression of doing something simple (the introduction of gpt can be felt also in this respect)

I like to maintain a friendly attitude with the company.

I liked the advice of building fast demos also expensive and then work to make them cheaper.

At week 4 i realized that writing this post was probably a good idea but i kept notes all way long.

By far the most useful part is to receive feedback from the user I dont think it was a smart choice to leave the field completely blank for us to assess, it was disorientating, but it was also fund

The impact of the mentor was huge from my side in order to understand on which analysis to focus on

It was funny to see that the clustering was not performed by them even if they are a digital company.

They do make a good job in maintaining the DB though.

WEEK 4 Preparing the slides to then receive the feedback 2 h before the meeting and going crazy. In the end during the meeting Maurizio was kind

WEEK 5 Still really amazed by code interpreter, not advanced code analysis, capacity Went all in into the Web behavio analysis and did a good job, check afterwards if what it did is sensible.

I wanted to do a streamlit app and it is easier than i anticipated. I wanted to create a big dataset with all the data before to load into it and then analyze but is best to just upload the single datasets and then do the processing gradually and then graph the results. Once again code interpreter helped a lot and it is useful to let it do his things giving the direction you want to go on rather than more instructions and then iterate on the bugs in order to improve the solution.

Now I am about to have a meeting with my tutor. I need to thank him because his direction was pivotal for me in order to THIS ARE NOTES DURING TH MEETING Rui is great, it is cool to see that siloutte and elbow methods agree. Also that i chose by hand 6 clusters and it was good I would say that it is important to name the clusters instead of leaving just names,

Comment Alessandro: we need to put it in Business Terms. What is the business terms proposal that we have? What to copy from Alessandro? –> Clearly engaged, asked questions during the presentation, directing towards what he knows its useful and he can help with. BUSINESS PERSPECTIVE –> Too many results wit In cluster 1 and 2 there is there are the users that invest the most. We should find how much the clusters overlap, quick win? Alessandro: Put all the information, population correlation, Take care of monitoring clusters 1 and 2, with no delay when their project end propose a new one How many people are there in any cluster ? How many projects did people invest on? Notice the importance of story telling that Alessandro keeps talking about It is very essential to be clear when communicating with people Rui proposes a follow up on the web behavior from my side.

The NUMBER OF PROJECTS DIVIDED BY THE TIME WOULD BE A GOOD IDEA TO CHECK Indicator to check the time ranges of projects. The easier that Alessandro can think of is user registration time INFORMATION ABOUT COMPANIES CAN BE TAKEN FROM PARTITA IVA THERE ARE “OPEN”/ PAID CONSOdata https://consodata.it/ —- Registro delle imprese https://www.registroimprese.it/ You can find potential companies. They can filter out t They data is also useful to map their custumer. You will find websites GOOGLE MAPS COMPANIES WITH RATING FROM PEOPLE AS A CLASSIFIER FOR THE COMPANY PREPROCESSING. or TRUSTPILOT but reviews can be fake. Results are the first thing if you have data science creation you start with Business Intelligence and then you build up. Rui Question: Is it useful to have a prediction of the campaign based on the info ? The weakness is a bit optimistic to pretend ChatGPT getting the info from it But you can use to recap the good points.

Then I need to present my results to Tiziano in order to speed up. I need to find critically what it might be useful and what not I think to start in cronological time in order to show him if he can pick up and improve some stuff or not. RUI NOTEBOOK IS A MUST.

Remember also to mention regarding the etoro comment that he can find in the slides the screenshot of the subscription procedure so that he can pick up on that.

WEEK 6 Once again yesterday I saw the needs of effective communications Rui was able to actually receive the feedback from the stakeholders of implementing the filter to type “crowd” but it was not really capable of of communicating to us (which we didnt understand neither me or Cristiano, and this necessity is something that emerges when parties are really busy, there communication is even more essential. ) Nor it was able to communicate to the stakeholders that the analysis was actually incorporating their requests, so that they complained to us. Even if he was always a step ahead of me, the focus of the conversation was centered around my presentation due it being simple to understand. There is a craft that i am growing of making complex things easy.

Incredibly important the shapeof your mind on a givne day. There are some days where you are just too tired and pushing too much is even detrimental if not useless. Somedays like this Tuesday here I got out of the house singing where in three hours i got out major results implementing the feedback it was given to me and expanding with good ideas for visualization.

Week 7 Interesting how the introduction of a data scientist in the mamacrowd team improved the communication from my side. It just became way more natural to exchange ideas and the work done. This could be a general insight in how actually in the composition of two teams that communicate is important the presence of a common link ( other than the management one? )

Also by the end, the ease of the work to be chosen is bigger. That is is much easier to understand what acutally is important to work on. This is good in many respsects. ———————————– Week 8 Molto interessente la storia di Translated In cui iniziavano con una webservice in cui un sito web proponeva il risultato di un traduttore umano vero (Isabell era la traduttrice, Marco il fondatore.) Interessante notare come questa visione è riproposta nel motto di pischool “Machine Intelligence Meets Human Creativity”

Nata nel 1999 è l’attività principale che ha permesso la creazione di Pi School da Marco Trombetti

Bellissimo girare attorno alla villa Pi6. Vero ambiente startup con macchinetta nespresso, villa stile giapponese, piscina e chupa chups gratis. Poi altra cosa divertente è il fatto che marco continua a comprare sempre più ville attorno al questo spazio e i vicini hanno paura di essere copmrati. Poi sempre in stile sempre italiano, la villa in cui siamo adesso è stata comprata dopo essere stata sequestrata da un politico per un’inchiesta di corruzione (probabilmente mani pulite)

Bello incontrare i compagni, sorpreso positivamente dal carisma dei miei compagni di corso. pErSoNaGGi Lo spirito è sempre quello di cercare delle persone stimolanti attorno.

Oggi 25 Settembre 2023 sono stato ufficialmente hackerato per la prima volta. La VM su cui lavoravo è stata bucata tramite mongoDB, ed è stato proposto un ramsowre