I have some `data` like:

    data = Uncompress[FromCharacterCode[
      Flatten[ImageData[Import["https://i.sstatic.net/1E5DB.png"], "Byte"]]]]

> `{{311.,191.},{324.5,374.5},{357.,436.5},{328.,730.5},{333.,1196.},{334.,1552.},{344.,1827.5}}`

If we use `FindAnomalies` then we can find a outlier:

    abponts = FindAnomalies[data3]

>`{{357., 436.5}}`

But I don't want to use `FindAnomalies` to do this, because it too slow and I don't know how to use other languages to imitate such neural network function.

The current idea is to fit a line for all points, then calculate the distance from the point to the fitting line. But it seems this distance is hard to distinguish that outlier. The red line is the fitting line in follow:

    Show[ListLinePlot[SortBy[data, Last], 
      PlotRange -> {{300, 1000}, {0, 2000}}], 
     Plot[Evaluate[Fit[data, {1, x}, x]], {x, 300, 1000}, 
      PlotStyle -> Red], ListPlot[data3, PlotStyle -> Blue]]

[![enter image description here][1]][1]


Could anybody can give me some advice?

  [1]: https://i.sstatic.net/lv4Jt.jpg