Regression between reflectance at pixel value vs input daily data over image collection

I have been trying to make a raster of regression values between reflectance value vs input data at each pixel position in a collection (So, for example the R^2 relationship between daily average precipitation and reflectence value at every pixel position within a given extent). I have tried doing this by filtering MODIS by a list of dates and applying a constant band to each image with a respective input daily data value from CSV data, but I am having trouble extracting a specific CSV value for an individual date and applying it to an individual, respective image. The following code has returned the error "single positional indexer is out-of-bounds" when using the python Google Earth Engine API :

    import numpy as np
    import pandas as pd
    import ee
    ee.Initialize()
    df=pd.read_csv('.csv')
    ROI=ee.Geometry.Polygon([coords])
    
    dates=df['Date'].tolist()
    df['DateMillis'] = df['datetime'].astype(np.int64) / int(1e6)
    data=df['Data'].tolist()    
    date = ee.List(dates).map(lambda date: ee.Date(date).millis());
    d=ee.List(data)
    
    def maskModisClouds(image):
        return image.updateMask(image.select('state_1km').bitwiseAnd(1 << 10).eq(0))
    
    def timeband(image):
         return image.addBands(image.metadata('system:time_start'))
    
    def databand(image):
        return image.updateMask(ee.Image(((df.loc[df['DateMillis']==image.get('system:time_start'),'Data']).iloc[0])).addBands(image))
        
    col = ee.ImageCollection('MODIS/006/MOD09GQ').filter(ee.Filter.inList("system:time_start", date));
    Image = col.filterBounds(ROI);
    image=Image.map(maskModisClouds).map(timeband).map(Databand)