Jupyter Notebook or Labs & Data Format


I’ve got 6 years worth of monthly sales data for 1,100 SKUs that I need to forecast. A couple of years ago I read an article on LSTM using jupyter notebook. I managed to download and install it and configure it and feed it my data (4 years back then) in excel format where each column had an SKU and each row had the 48 monthly data points for each SKU.

Now Im back to my old ways and read back over my notebook to see what I did. This time the data wasnt provided to me by my purchasing assistant but rather I had to mine it and found that the original format is transposed. So SKUs on rows and monthly data points in columns to the right. I know transposing is easy but I want to know:

  1. Is there a way for me to use the data in the current form and save myself, not so much the work but avoid silly paste mistakes?

But more importantly i ran into an article that mentioned how jupyter labs is so much better than notebook. I read the reasons but I dont remember so clearly that they convinced me. I guess at first it struck me as awesome because i somehow thought labs was more complete and therefore smarter and I could just feed it the data and maybe config the parameters in a GUI assisted by some “AI Wizard” and help me find patterns as well as simple forecasting. But I think I remember most advantages were just easier workflows or something like that.

  1. Does anyone recommend jupyter labs over notebook? To put things into context im a general manager that every so often dives into sales with the team but then have to meet with purchasing or finances, so I dont have an endless amount of time to immerse myself in data mining and analytics. Although I do have a scientific-mathematical background.

In the end I want the simplest option to feed the data and find patterns with the aid of AI and get some practical forecasting done and every now and then go back and peruse the data to find extra tidbits.

Im attaching my notebook of what I did last time. Ok i guess im not, i couldnt attach any file types other than image files I guess.

JupyterLab (installed as jupyterlab) and the classic Jupyter Notebook (installed as notebook) are both browser-based client-server tools that can both open .ipynb and text files, start terminals, etc… As-installed, they make no use of any AI/ML techniques.

From an installation perspective: I’d recommend installing jupyterlab, as it will bring notebook along, and you can try them both and see which one is better for your task.

Both can be extended, however, but I don’t know of a drag-and-drop-and-get-free-answers-to-arbirtrary-questions extension for either. More likely, these will happen in the underlying kernel you use (likely ipykernel, but it’s hard to tell from your description), and will work independent of which editor you use.

A quick web search suggests that pandas sklearn timeseries forecast will give you a lot of concrete write-ups to peruse: here’s the top hit.