[Day 160] Simple data engineering pipeline with Prefect, and... MLOps with mage.ai (tons of problems)

 Hello :)
Today is Day 160!


A quick summary of today:
  • simple data engineering pipeline with prefect
  • tons of trouble learning about orchestration with mage.ai


After yesterday's journey with prefect the youtube algorithm recommended me another tutorial for prefect - this time for creating data pipelines with prefect. So I decided to give it a go. 

What is data engineering?

  • data scientists can do data engineering, but in specific cases where the two jobs cannot or are not needed to be separate
  • data engineers build databases, they build lots of data pipelines and manage infrastructure (also care about cost, security)
What are data pipelines?
  • ETL(ELT)/batch pipelines that move data from A to B
    • databases, APIs, files
  • streaming pipelines - as data comes in, we consume that data and send it wherever it needs to go 
    • message queues, polled data
The main github repo used is here.

After some basic setup, when we run 'pipeline' in the terminal which runs the main.py file:

In prefect we get a flow run

and logged outputs from the petstore url

Now for a flow with a bit more tasks~

1st task: retrieve from API

2nd task: clean data

3rd task: insert to postgres db

final flow:

Success!
We can also see the data loaded into postgres:

Beege (the teacher) showed also how to do simple task tests and avoid a common prefect error when executing prefect task functions outside a prefect flow



Prefect is amaing, but the MLOps zoomcamp course works with Mage.ai this year, and after I gave up on it due to errors that I could not fix last time - I decided to give it another go today. And... omg so many errors and problems - I almost gave up on it several times. The videos are recorded 2 weeks ago, and yet there is so much different on my UI (and other students' UI) that it is mind-boggling.

The content was not much, but the constant errors... made me invest upwards of 12 hours of bug fixing for a total of 30 minutes of video content. 


Orchestration with Mage.ai from MLOps zoomcamp module 3

The content of the intro to orchestration is:

I have to do 3.5 tomorrow (hopefully). 
I just want to say a huge thank you to the QnA bot on the mlops zoomcamp slack channel that at least pointed me into the right direction to solve my issues. Just quickly - I am bit sad that there are so many issues (almost every video) because it will drive people away from this amazing course, and below I will just provide my successes, and if you are really curious about my problems - they are all on the datatalks slack channel haha. 

I started from the beginning

3.1 Data preparation

I created the above pipeline that reads NY taxi data, does a bit of preprocessing, and then outputs X, X_train, X_val, y, y_train, y_val, dv. By the way, each block is a piece of code.

3.2 Training

First, create a pipeline to train a linear regression and lasso model. It takes data from the above 3.1 pipeline, loads models, does hparam search, and finally trains the two models.

Secondly, I had to create a pipeline for an xgboost model
The last (pink) bit is connected to creating visualisations. 

3.3 Observability

We can create any kind of bar/flow/line/custom charts, and the above 3 are some SHAP values from the xgboost model. 

3.4 Triggering

Here I created an automatic trigger to train the xgboost model when new data is detected.

Also, created this predict pipeline

There is a nice way to setup a basic interface for inference

We can also setup an API to do inference through that as well:

That was all for today. I have very strong feelings towards Mage.ai but for now I will roll with it because of the nice course. Otherwise I will try to use prefect for some side project (have to think about that a bit). 


That is all for today!

See you tomorrow :)

Popular posts from this blog

[Day 198] Transactions Data Streaming Pipeline Porject [v1 completed]

Image
 Hello :) Today is Day 198! A quick summary of today: data streaming pipeline project [v1 done] Here is a link to the project's repo. Well ... I did not know I can do it in a day (~14 hours) after yesterday's issues but here we are. Turns out in order to insert the full (~70 variables with nested/list structure), I need the proper pyspark schema. And yesterday I did not have that and that is why when I was reading data in the kafka producer I was getting NULL in the columns - my schema was wrong. Well today I not only fixed the schema for the 4 variables I had yesterday, but included *all* the variables that come from the Stripe API ~ 70 (for completeness).  When I run docker-compose, the data streams and is input into the postgres db (and is still running). Unfortunately, the free Stripe API for creating realistic transactions has a limit of 25, so every 3 seconds, 25 new transactions are sent to the db. It has been running half the day (since I got that set up) and as I am wr
Read more

[Day 107] Transforming natural language to charts

Image
 Hello :) Today is Day 107! A quick summary of today: created text2chart - transforming natural language to charts [ github ] [ webapp ] I am super sick, but in the around 3 hours of my fever not being high, I managed to study a bit. I was again thinking about creating some nlp project but was stuck on the idea of how to host a model. After looking around I found this article where they create a similar app called chat2vis  and they use codellama. I know codellama is an open source model that I can get from huggingface and I wondered, how did they host it. Turns out - they ask the user to input a huggingface api key, which then allows for querying and generating content. (AWESOME! this is applicable to some of the other PDF chat apps that I have, but I will look to make an app of them later).  My fever eased a bit, and I sat on my chair to do some work ^^ Here is a breakdown of the app. I used streamlit. Also, 2 default datasets during dev: Financial Statements of Major Companies(2009-
Read more

[Day 54] I became a backprop ninja! (woohoo)

Image
 Hello :) Today is Day 54! A quick summary of today: Read through cs231n's neural network lectures [1] [2] [3] Spent 8 hours going over Andrej Karpathy's backprop ninja tutorial After yesterday's confidence boost in doing backprop, I came in today with the intention of becoming of, as Andrej Karpathy calls it, a backprop ninja.  The cs231n NN lectures were the usual about how NN became to be, param initialization, activation functions, loss functions, regularization and optimizers. Nothing out of the ordinary, but still was good to reassure myself of that knowledge. But that was definitely not the main even of today. As I mentioned, I wanted to become (or at least attempt to become) a backprop ninja today. I have watched that particular tutorial twice, the 1st time I had no idea what is going on, the 2nd time I could follow, and now it was the 3rd time - and now, I wanted to participate and actually see the why and the how of the things we have to do, when we do manual back
Read more