.png)
Shipped this project!
Today, I shipped a full-stack quantative modeling platform focused on temporal pointp processes - specifically implementing and comparing Poisson and Hawkes processes from scratch.
Shippeddd
Validated
On to the next layer
Activity
Hola amigo guyss,
Yayyyy!!! Lessgooo I finally finishedd the project
It was pretty difficult tho
The last part was basically fixing and adding the backend of the website and linking it to frontend
Also wanted to debug the code a little bit so did that
It was fun thooo
GOOOO CHECKKK ITTT OUTTTT!!!!!!!!!!!
Thankkk youu!!
0
Log in to leave a comment
Hola amigos,
This might be one of the last update on this project coz i just deployed my project, Completed and debbuggeddd the code fully
The only thing that can be more interesting it the content in it,
ill try to make it more intereseting
https://lambdalab.vercel.app/ is the link to the deployed website
I accept i did use ai , but remember it has limits, so i had to go through the code in order to check if it was working or not..
(just noticed for some reason the deployed demo isnt working so i might have to fix that :C )
Anywyas do lmk your opinions in the comment box
Anyways thankk youu
0
Log in to leave a comment
Hola amigos,
I dont have a lot to show today , but most prolly i have done lots of coding today, and almost finished all the backend and i have nothing more to show.
Next task is to create ai prompts for the front end
i almost designed how it should look like
Below will be the figma designs
Bbye till i see you next time
0
Log in to leave a comment
Hola amigos,
So this one was mostly my learning phase , where i wrote all my wrotes by myself (USED GPT TO UNDERSTAND!!!)
Mainly covered
- Poisson Process
- Hawkes Intensity
- LogLikelihood and some more mathematical concepts that were required for this project
Moreover
i also made a simulation for Poissons process and Hawkes Process in order to compare both of them
It was pretty fun
Andd sorryy got prettyy lost of time coz i was studying the concepts…!!!
do lmk if any changes should be made
0
Log in to leave a comment
Heyy Amigos,
Great to see ya’ll….
The first step towards a project is to understand what are we building , so this is what i did today ,
I basically tried to understand what am i actually thinking and wrote it down
Nothing special
- This included the updation of github’s README (id : 03924b7)
Nothingg crazzzy as in such lol
Please consider following me - that motivates me a lot and i really mean it
THANKKK YOUU .. see you nextt timeee
0
Log in to leave a comment
So basically i fixed , edited and hosted the XGBoost using streamlit , nothing special , i just know how to make models , so it was kinda new experience to me , lmk if any changes
0
Log in to leave a comment
I agree i used AI to design the front end , but even that requires skills :D
Anyways here are some updates on that
0
Log in to leave a comment
Shipped this project!
So this is a rawdogged and partially self-made XGBoost implementation , the thing that i find interesting is the mathematical importance behind it , as a math undergrad , i hope even you find that intereseting , the initial devpost just shows what and how does the math work here , it is really interesting please do check out , and let me know what should i improve coz i am here to improve and not to get the prizes only :D , i hope you really love it …. THANKSSS T_T
We brought machine learning :
Loaded the data set and got a feel for all the ingredients.
Split it into training and testing, keeping it honest.
Fired up an XGBoost classifier with a binary:logistic objective.
Trained the model, then made predictions on both labels and probabilities.
Tasted the confusion matrix for errors.
Tuned hyperparameters with GridSearchCV to spice things up.
Chosen the best model and checked it on unseen data.
It is running on enhanced trees that really bring the heat.
Dont judge my screenshot , i wasnt sure abt what to put
0
Log in to leave a comment
So this isnt any technical stuff or someshit.. LOL … Here i just wanted to discuss the mathematical challenges and how did XGBoost work
- It is basically based on trees and depends how are we gettting the leaves of the trees
- we calculate Output and loss for every tree using calculations(formula based)
- It has manly 2 types XGBoost Regression and XGBoost Classification and both are almost similar , just have minute differences in the formulas
0
Log in to leave a comment
In real-world traffic networks with multiple nodes and multiple possible routes, individual drivers choose paths that minimize their own travel time.
However, these individual choices often lead to inefficient global outcomes, increasing overall congestion and total system travel time.
The challenge is to model, analyze, and influence route selection in complex traffic networks so that the system performs better without forcing centralized control.
0
Log in to leave a comment