F1-Score provide a balanced measure between precision and recall.

True Negative Rate (TNR) - is TN/TN + FP

True Positive Rate (TPR) - is TP/TP + FN

Lately I was in to cyber security when I create this channel but now I will be posting on ML/AI continuously.

Have you taken this test https://personalitytests.com/riasec/

If you ask me why I named my channel 'investigator ' here is why ... Good Night 🚶‍♀️‍➡️

#Opportunity_Alerts📣

🌍 CALLING ALL TECHIES, ACTIVISTS & CHANGE-MAKERS ACROSS AFRICA!

💥The Sub-Saharan Africa Digital Democracy Virtual Hackathon is your chance to be part of a movement building digital tools that protect rights, amplify voices, and empower communities. This exciting regional hackathon is part of the global Digital Democracy Initiative, bringing together the brightest minds across Africa to imagine and create tech for social impact.

Build solutions that:
🔸Strengthen governance & civic participation
🔸Protect activists and digital rights
🔸Combat disinformation, surveillance, and censorship
🔸Empower marginalized voices & communities

🗓9–10 May 2025 | Virtual
🎯Ages 18–35 | Teams must include at least one female member
💰5 winning teams will receive a US$ 7,500 grant to implement their idea.🏆

🔗Apply: https://hackathon.magambanetwork.com

📝Deadline: 30 April, 2025

"If this isn't for you, please share it with others who might be interested."🙏

Follow us👇for more opportunities
@opportunity_alerts

I lost my strike yesterday 😭 ... FYI it is in kaggle

I spent my morning learning data science, then attended class. After that, I had basketball training and returned home feeling very tired at 8 PM. Despite not being in charge of designing our hackathon website, I spent sometime with the design. Eventually, I went to sleep since I needed to travel to my town today. In all the hustle, I forgot to edit my notebook or do anything on Kaggle. My 17 days consistency is gone 😕.

Happy good Friday!

After the holiday work, I got back to my project. I was struggling to use 3D animated models with React, but then I watched a video and realized my 3D model was missing something important. So, I spent hours searching for tools to generate 3D models with rigs. Unfortunately, the free tools were really limited when it came to animation. In the end, I decided to start learning Blender. Now I have cute cat model but rigging is super tricky, the bones aren’t aligning well.

My point is hackathons are great. You’ll learn a ton.
I know it’s late, but Happy Easter! 😊

After spending countless hours diving into 3D animation and experimenting with Blender ( which is needed for my project) , I wanted to share some key takeaways to help anyone who’s just getting started. If you’re an aspiring 3D artist, game developer, or just curious about how all this works, here’s a roadmap that’ll save you time and frustration.

1. Start with Image-to-3D Model Tools

Head over to Ideate.xyz — it gives you a limited number of free tokens to generate 3D models from images. It's not perfect (you’ll need to polish the results), but it's a great way to quickly get a base model.

2. Polish in Blender

After generating your model, take it into Blender. This is where the real magic happens — you can clean up geometry, fix texture issues, and prepare the model for animation.

Not sure how to do all that? You already know the drill: YouTube is your best friend. Tons of tutorials out there will guide you step-by-step.

3. Use Pre-made Models (To Save Time)

Not everything has to be built from scratch. Here are some great resources for high-quality, free models:

- Models Resource - Coraline Character
- Sketchfab – Browse thousands of free downloadable 3D models, some even come with animations.

4. Add Animations with Mixamo

Once you’ve got your character, head to Mixamo . It has a variety of motion-captured animations (mostly human-focused). Before using Mixamo, your model needs to be rigged — think of this as giving it a skeleton with joints so it can move.

For automatic rigging, try AccuRig — it adds bones and joints to your model, including fingers!

5. Add Facial Expressions (Blink, Smile, etc.)

Want your character to blink, smile, or talk? You’ll need to use shape keys in Blender. Again, YouTube tutorials are super helpful here. Just note: this step requires a well-modeled face for good results.

6. Integrate into a React App with Three.js

Once you’re happy with your animation and model, you’ll want to bring it into your web app. Here's how:

- Convert your model to GLB format.
- Use React Three Fiber to display it in your React project.
- Use this amazing tool: GLTF.pmnd.rs – it helps you visualize your model and gives you the React code you need to embed it.

---

Finally

After going through all this, I realized how much time and effort it takes just to create and animate a single motion. It’s intense.

As an aspiring machine learning student, I’ve started a side project combining 3D animation with ML — and it’s shaping up to be something exciting. Feel free to reach out if you’re working on something similar!

//credit for chatgpt for organizing my post in good way 

Good night ppl🚶‍♀️‍➡️🚶‍♀️‍➡️

It's time to talk about where to host your Python apps! 🚀 

Render → Slow but amazing! It has database limits, but you can use your own custom domain. Supports hosting Django, Flask, FastAPI, and even Telegram bots. The main drawback? It's slow—because it's free. 
Vercel → Uses serverless functions to host Python frameworks, making it faster than Render. The main drawback? You can't write to files or have any database writes. However, temporary files and external databases are possible. 
Railway → Solid choice. 
Koyeb → Never use this for Telegram bots! 
Back4app → Use Docker for hosting. 
PythonAnywhere → Another option worth considering. 

deepseek is reasoning in Chinese😂 my prompt was in english looks like it prefer to reason in his mother tongue 🤪

https://youtu.be/-7AeNKev14M?si=5iXP91B1obdLfOrm Interesting😄

As much as I love learning, I have a bad habit of starting and not finishing most courses. So, to break that cycle, I completed the Computer Vision course from Kaggle today!
Btw if you want to learn machine learning Kaggle offer free courses from basics of programing to advanced ML topics.

Found this service by devin.ai that breaks down and lets you talk to a public Github repo. According to the site description its "Deep Research for GitHub".

https://deepwiki.com/

#Github #AI #Repo

From CEO of OpenAi 🤌
https://www.youtube.com/watch?v=5MWT_doo68k&list=PPSV

It's fascinating to hear about the future , safety and security of Ai from CEO of OpenAi. and some gossip about whether DeepSeek have affected them or not. 😁
if you care about Ai , I recommend you to watch it

Today I found this library while trying hand tracking project , check it out for your Ai project
https://ai.google.dev/edge/mediapipe/solutions/guide

https://colab.research.google.com/github/spmallick/learnopencv/blob/master/Introduction-to-MediaPipe/MediaPipe-sample-solutions.ipynb open source code to see how to use

I was training credit card fraud detection model with random forest .
The above image is confusion matrix which is evaluation metrics and can see the model has good performance over test data with small false positive ( classifying fraud data as non ). The data I used is cleaned data you can practice hyper parameter tuning. https://www.kaggle.com/datasets/nelgiriyewithana/credit-card-fraud-detection-dataset-2023