Explore possible patterns with a time series data
Budget: $750 – $1,500 USD
The job is to train an AI model based on a time series dataset of ~4000 data points. The data is just a list of 3-digit number in temporal order. Note that the first number is the latest and the last number is the earliest. So you might want to flip the order first.
The goal is to create a model that takes a few numbers (with size of your design) as input and spit out the next number as a prediction, with probabilities.
We will reserve 10% of the numbers as validation. Any model that gives an accuracy better than random guess will unlock payment milestone 1. If the model can reach 25% accuracy, you will get the full payment.
Note that the order of the 3 digits of the output does not matter.
For example:
[189, 297, 335, 108, 781] ==> [839]. Here the prediction 839, 893, 938 ... are all considered correct.
The goal is to create a model that takes a few numbers (with size of your design) as input and spit out the next number as a prediction, with probabilities.
We will reserve 10% of the numbers as validation. Any model that gives an accuracy better than random guess will unlock payment milestone 1. If the model can reach 25% accuracy, you will get the full payment.
Note that the order of the 3 digits of the output does not matter.
For example:
[189, 297, 335, 108, 781] ==> [839]. Here the prediction 839, 893, 938 ... are all considered correct.
Related categories:
Data Processing
Algorithm
Machine Learning (ML)
Statistical Analysis
Neural Networks