Discover your perfect match with AI-ML powered Vedic astrology for accurate marriage compatibility predictions.
Boasts a 90-95% accuracy through deep learning based on 100K+ Kundali Matchings researched and validated by AstroNidan team.
5 |
|
59% |
4 |
|
31% |
3 |
|
9% |
2 |
|
1% |
1 |
|
0% |
Prediction | |
---|---|
Accuracy | |
Readability |
Boasts a 90-95% accuracy through deep learning based on 100K+ Kundali Matchings researched and validated by AstroNidan team.
You will get answers to the following -
The traditional Ashtkoota milan or Guna milan method, based on 36 points, has often been insufficient for accurately predicting the success or failure of a marriage. This approach has resulted in the rejection of many good matches and the acceptance of bad ones. Our marriage compatibility product addresses this issue. Thanks to advancements in AI and advanced data analytics, we have developed a holistic model that takes into account planetary positions, houses, doshas, Ashtkoota, and their exceptions to make decisions about Kundali Matches. Our system provides decisions at three levels: 'Good to Proceed,' 'Proceed with Remedies,' and 'Try to Avoid,' along with a score from 0 to 10 for easy comparison.
Classical astrology books such as Brihat Parashara Hora Shastra, Uttara Kaalamrit, Bhavarth Ratnakar, Sarvarth Chintamani, Laghu Parashari, and Madhya Parashari provide various principles for predicting compatibility. However, applying these principles using the human mind can lead to a lot of errors. For the sake of simplicity, only the Ashtakoota Match has been very popular, which has in turn led to many errors in Kundali Matching.
In AstroNidan, we have programmed thousands of logical statements in Python language and trained and validated them through multiple AI/ML models over 100,000 Kundalis for best profession prediction.
Some of the salient features are -
We have a team of Data Scientists, ML Scientists, Data Engineers, Sanskrit Scholars, Astrology Researchers, and Python Programmers engaged from top universities (IIT, IIM, JNU & BHU), who have worked tirelessly to build this product.
We have prepared this report after factoring in -
We have used semi-supervised learning methods to explore hidden patterns & combinations and evaluated numerous ML models such as Neural networks, SVM, Random Forest, and gradient Boosting algorithms. We shortlisted the best-performing ML models and fine-tuned them further through our robust feature selection process.