Python code for estimation value in dataset
Budget: $250 – $750 USD
I need someone who is very good at machine learning and data analysis and has proven experience.
For this project, I have various crypto assets which has prices split into 3 minute epochs of data (The data is taken from lots of exchanges)
Each epoch has 3 minutes of data and various companies (including myself) submit what they think the true price of the asset is for the period in USD, so might take weighting average of certain price times etc.
Once all companies submit their estimate of the true price, the external service then calculates the median of prices submitted and a quartile range. My submitted price needs to be inside that quartile range as often as possible for my service to be considered accurate.
Attached are 1000 sample epochs of data for each of the crypto assets (11 in total). In these files, they have the prices for each time for each exchange and also the upper/lower quartile range.
To score accuracy, generated prices are round to 5 decimal places and if inside the upper/lower quartile range, gets 1 point.
If exactly the lower or quartile range price, 0.5 point. If outside, 0 points.
0.6 average points or above over all assets and epochs would be a good result and would be what I'm aiming for. If possible, higher.
So I need ML learning to be applied to analyze the data and predict what that price will be.
For this project, I have various crypto assets which has prices split into 3 minute epochs of data (The data is taken from lots of exchanges)
Each epoch has 3 minutes of data and various companies (including myself) submit what they think the true price of the asset is for the period in USD, so might take weighting average of certain price times etc.
Once all companies submit their estimate of the true price, the external service then calculates the median of prices submitted and a quartile range. My submitted price needs to be inside that quartile range as often as possible for my service to be considered accurate.
Attached are 1000 sample epochs of data for each of the crypto assets (11 in total). In these files, they have the prices for each time for each exchange and also the upper/lower quartile range.
To score accuracy, generated prices are round to 5 decimal places and if inside the upper/lower quartile range, gets 1 point.
If exactly the lower or quartile range price, 0.5 point. If outside, 0 points.
0.6 average points or above over all assets and epochs would be a good result and would be what I'm aiming for. If possible, higher.
So I need ML learning to be applied to analyze the data and predict what that price will be.