Don't worry too much about missing data
Рет қаралды 738
Күн бұрынMissing data can be highly annoying to deal with. Some algorithms can't deal with it, so your first instinct might just be to impute the values. It turns out though, that sometimes you're able to 'ignore' these missing values. It all depends on the machine learning algorithm that you're interested in using.
00:00 Missing values
02:12 Imputing Pipelines
04:11 Alternative
05:06 Whiteboard
07:50 Reflect
@DogTennisFreak 21 күн бұрын
Excellent video, love the thorough walkthrough and diagrams!
@probabl_ai 19 күн бұрын
Glad it helped!
@volodymyrholubets1872 23 күн бұрын
Brilliant talk, Vincent.
@probabl_ai 23 күн бұрын
Happy to hear it. Any requests for future videos?
@jonathandawson3091 23 күн бұрын
May be short case studies? And modeling on datasets using different techniques? I personally would like about 5-15 minutes videos (though not sure about others - models may be created once you have enough videos to predict what people watch more!)
@volodymyrholubets1872 23 күн бұрын
@@probabl_ai yes, please make a video about dimensionality reduction and visualisations. (UMAP in particular)
@jonathandawson3091 23 күн бұрын
I appreciate this video very much. God, I miss sklearn.
@probabl_ai 23 күн бұрын
No need to miss it, because it has never gone away
@joaomurilopalonefauvel2843 23 күн бұрын
I have data from a meteorological station with missing values at night due to the battery running out. Any suggestions on how to deal with that?
@CoderShare 22 күн бұрын
Get a bigger battery.
@probabl_ai 19 күн бұрын
It might depend on the usecase for this data, if you never care about making predictions when you're out of battery then it might be fine?
Wrecker
Рет қаралды 567 М.
Rj Mobile 01
Рет қаралды 14 МЛН