How to compute a zonal statistics in timeseries images in Google Earth Engine?

Question

I have computed NDVI for January 2018 to December 2018 from Landsat images that give 21 NDVI bands. I have a shapefile with three polygons; I would like to have three time-series NDVI values corresponding to the polygons in the given shapefile.

Imports (2 entries) var L8: ImageCollection “USGS Landsat 8 Collection 1 Tier 1 and Real-Time data TOA Reflectance” var roi: Table users/user_name/sample_shapefile

// NDVI function

  function addNDVI(image) {
    users/user_name/sample_shapefilevar ndvi = image.normalizedDifference(['B5', 'B4']);
      return image.addBands(ndvi);`
       }
// Date range 2018 jan to dec

   var filtered = L8.filterDate('2018-01-01', '2018-12-31')       
.filterBounds(roi);` 

var with_ndvi = filtered.map(addNDVI);

    var ndvi = with_ndvi.select('nd');

var dt = ndvi.reduceRegion({
  reducer: ee.Reducer.mean(),
  geometry: roi.geometry(),
  scale: 30,
});
print(dt)

I would like to receive three time-series NDVI values one for each polygon in the given shapefile.