Data Science - Expectation vs Reality (funny!) - ft. @KenJee_ds

first day on the job super excited to learn about my project hey mr g hey luke i i told you before you can just call me ken okay mr ken okay i sent you over an email with some of the project requirements and documentation did you get a chance to review it yet yeah the project outline is really well detailed it's even better than what some of my teachers provided in school for our projects thanks luke i appreciate that we really take pride in our documentation here so just a quick overview of the project so first we're going to clean up the data a little bit i'd love for you to perform at eda next we're going to want you to go through and do some model building finally you'll present these findings to all of the stakeholders here yeah super simple perfect you have any questions no not at all thanks mr ken first day on my job super excited to learn about my project hey mr g how's it going what uh did you get that email that i sent over with the with the project overview actually i haven't received any emails um maybe maybe i didn't send it uh i'll just give you a quick overview of the project is does that work for you okay so all right first does that make sense does that sound good yeah actually i do have a few questions so far i got to run to another meeting but bye time to get into cleaning the data let's connect to the database that was so easy to connect let's see what the data looks like i'll just do a quick little query wow that's some pretty clean data already well almost clean it looks like a few columns still need to be cleaned up okay now let's see what this newly cleaned data looks like wow that's some clean data time to clean the data let's connect to the database wait mr g didn't provide me the login information for that i'll just call and ask what have you cleaned that data yet um no sir i don't think i have access to the data yet uh yeah yeah just check with i.t they'll get you sorted out man who would i see do a contact gotta go uh you know work calls um bye i guess i'll just message them on slack heck yeah i'm killing this what all right time to get into some good old [Music] eda that was so much fun oh time to meet with ken so ken the major insight that i gained from my eda was that if you have a papaya and you apply 300 newtons of force and no indentation is left that is the perfect ripeness and that's going to result in an exponential increase in sales honestly luke this is this is groundbreaking this is going to set you up perfectly for designing the papaya sales model ah thanks ken finally got access to the database time to get into some good old eda ha drop definitely don't want to delete the database no ah where is it [Music] oh hey mr g um you look great today luke what is going on man i just got an alert that said our app lost connection to the database well what happened is i don't care what happened just fix it now yes sir i knew i should have hired alex freeburg for this what what nothing nothing i am so fired all right on to building the sales model let's load in some of my favorite libraries okay libraries are in let's start building our model a little later looks like training is complete for my first model let's run some initial tests on this i remember 70 to 90 being a good accuracy score holy smokes 95 and precision wow 93 heck yeah let's look at auc of the roc wow that's around 90 as well man this is awesome i love my job recovered the database through eda now time to build a model let's load in some of my favorite libraries three hours later rookie mistake need to pip install first let's split this data set into a training and test set any unbearable hours later all right first model is built let's check out some of these scores 46 that's worse than flipping a coin let's check precision 38 even worse okay what does roc look like area under the curve of less than 50 percent is that even possible i'd be better off flipping a coin to predict this through my analysis not only are we able to grow the papayas in a more stable manner to prevent the ring spot disease but also it's predicted to grow our sales by over 150 percent luke man you really outdid yourself on this one the engineering team they're gonna be implementing this in the next two to three days we're getting this out of the door quick because it's driving so much value ah this is awesome to hear ken but honestly i can't take all the credit pandas did most of the heavy lifting luke now that that everyone's left the meeting i'm i'm here to give you the even better news we've decided to promote you we've also decided to give you a fairly sizable bonus for your contribution in this first week ah thanks ken although slightly behind schedule some below average model scores and the papayas are still being savaged by that ring spot disease we were able to achieve with my model a sales growth rate of 3 percent which mind you is more than inflation luke can you hear me ken here uh i know you spent a lot of time a lot of time on this and uh unfortunately based on the model performance we're gonna have to go just a little a little different direction with this project um what does that mean yeah based on stakeholder alignment uh a lot of the the data interpretation and business need we've decided to go with with a different model i just invested six months of my life into this project are you going with alex's model yeah just better luck with the next model let's let's uh try to to keep things more organized and uh and really understand the business case better what the you

Channel: Luke Barousse

Views: 42,828

Keywords: data viz by luke, business intelligence, data science, bi, computer science, data nerd, Ken Jee, expectations vs reality, data science projects, data science sketch, data science funny, evp vs reality, funny, luke & ken, Ken & luke, data analysis, EDA, project brief, ML model, Model build, ml model build

Id: 78LMtFItAdw

Length: 7min 38sec (458 seconds)

Published: Fri Aug 06 2021