Instantly see value of AI and data science through real-world demos. You will get the answer to "WHY" AI and Data science through practical examples. All videos are code-free , so that you can understand the value. You can also try out the demo on my platform experiencedatascience.com, with your data.
Interesting demo but forgive me I isn’t it just a fancy “look up” table ? I mean the predicted price is just checking if a descriptor is im a table or not and then applying a price accordingly? Apologies if I’ve missed the insight here. Appreciate any clarification cheers
Hi, Thanks for your comment. The look up table technique does not work as the items which are put in sale have a description which is given by a seller in free-form. Every item description is generally unique even if it’s the an identical product. Also sellers put many specific things and personalize the description. So the exact description in not available in past sales . Hence it’s required to use a LLM approach Hope this helps . Thanks for watching my channel
I can see using LLMs to establish better embeddings to run the traditional pricing algo's (time-series, regression, decision-trees off), but it's not going to give you optimised elasticities on its own. Unless I'm grossly mistaken.
Hi, once you are on my website (experiencedatascience.com/), please login. Once you are logged in, go to See All Experiences. Then select Data science and data analysis. Then you will see the "Health Activity Analysis" which corresponds to the project in the video. Let me know for any questions. Thanks
Good job I liked the demonstration, I think you should further explain the hyperparameter tuning in dbscan because it can drastically change the results
Hi sir, I didn't really understand the cluster analysis - what are the different colours, what is the trend, what do the different colours represent? Thank you!
Hi, the cluster groups similar reviews together. The colors signify reviews of similar products. For example at 1:51, the red cluster on top left of the screen is related to dog food
Hi, Let me check. In the meantime, you can also do same analytics as follows - 1. Use menu Datasets-Play Datasets to copy taxi_data_porto_location dataset. 2. Then select Datasets-Your Datasets, select the taxi_data_porto_location, and select Analytics. You will see all analytics including histogram, boxplot, geolocation etc..
Thanks for the video. In the 2-dimensional plot, we have reduced the 1540 vectors to a 2-d in-order to be able to plot them. Which algorithm did you for this reduction? t-SNE, UMAP, or some other algorithm.
@@DataScienceDemonstrated I don't see any models mentioned in the description. I would be expecting gpt-4, gpt-3.5-turbo or any other models OpenAI provide. It would also be great to add the prompt used to get the sentiment.
hi, thank you for the vivid explanation. may i ask a question: which software are you using to group different product items into clusters, and then visualize those clusters with color on the x,y coordinate?
Thanks! I have created my own platform , which is based on Python and JavaScript visualization libraries. You can access it here : experiencedatascience.com . You will be able to make similar clustering and visual as I have shown, without coding. Hope you enjoy it
Great you liked it and thanks for the feedback. You can try to see my medium blog link, which has technical implementation of similar sentiment analysis , but with hugging face models .
