label_to_index

        label_to_index,

    confusion_matrix =
        MLBase.confusmat(c, actual, pred)
        #freqtable(DataFrame(targets = actual, predicts = pred), :targets, :predicts)

    confusion_matrix = MLBase.confusmat(c, actual, pred)
    #freqtable(DataFrame(targets = actual, predicts = pred), :targets, :predicts)

        @time train_accuracy, train_confusion_matrix = evaluate(model, train_sample, classes)

        @time train_accuracy, train_confusion_matrix =
            evaluate(model, train_sample, classes)

            #a = test_confusion_matrix[1,1]
            let _model = cpu(model)
                jldsave(
                    "model_$(model_name)_CPU_epoch-$epoch-$test_accuracy-$(today()).jld2";
                    model_state = Flux.state(_model),
                )
                @info "Saved a best_model"
            end

        #a = test_confusion_matrix[1,1]
        let _model = cpu(model)
            jldsave(
                "/media/david/SSD2/model_$(model_name)_CPU_epoch-$epoch-$test_accuracy-$(today()).jld2";
                model_state = Flux.state(_model),
            )
            @info "Saved a best_model"
        end

glob_pattern = "Clips_2024-06-26/"

glob_pattern = "Clips_2024-10-21/"

function resample_to_8000hz(signal, freq)
    signal = DSP.resample(signal, 8000.0f0 / freq; dims = 1)
    freq = 8000
    return signal, freq
end

model = load_model('/home/david/best.model')

model = load_model('/home/david/best.model0')

folders =  glob('./*/*/')

folders =  glob('./*/2024-10-18/')

    scores.to_csv("scores-2024-08-29.csv")
    preds.to_csv("preds-2024-08-29.csv")

    scores.to_csv("scores-2024-10-21.csv")
    preds.to_csv("preds-2024-10-21.csv")

#=Kahurangi
folders =  glob('./*/')
for folder in folders:
    os.chdir(folder)
    print(folder, ' start: ', datetime.now())
    # Beware, secretary island files are .wav
    field_recordings = glob('./*.[W,w][A,a][V,v]')
    scores, preds, unsafe = model.predict(
            field_recordings,
            binary_preds = 'single_target',
            overlap_fraction = 0.5,
            batch_size =  128,
            num_workers = 12)
    scores.to_csv("scores-2024-10-21.csv")
    preds.to_csv("preds-2024-10-21.csv")
    os.chdir('./..') # Be careful this matches the glob on line 284
    print(folder, ' done: ', datetime.now())
    print()
    print()
=#

# needs SSD1 present for dawn_dusk.csv

folders=glob("*/2024-06-23/")
for folder in folders[2:end]

folders=glob("*/2024-10-18/")
for folder in folders[3:end]

        CSV.write("/media/david/SSD3/New/pomona_files_20240627.csv", df; append=true)
    catch
        @warn "error with $folder"

        CSV.write("/media/david/Pomona-4/Pomona-4/pomona_files_20241018.csv", df; append=true)
    catch x
        @warn "$x error with $folder"

cd("/media/david/SSD3/New/")

cd("/media/david/Pomona-4/Pomona-4/")

Then go into sublime and change drive New to Pomona-4

COPY pomona_files FROM '/media/david/Pomona-3/Pomona-3/pomona_files_20231019.csv';

COPY pomona_files FROM '/media/david/Pomona-3/Pomona-3/pomona_files_20241018.csv';

To restore from backup:
duckdb my_database.duckdb
IMPORT DATABASE 'AudioDataBackup_2024-07-10';

predictions = glob("*/2023-09-11*/preds*")
predictions = glob("path/to/preds*")
for file in predictions #[1:6][7:12][13:18][19:24]

predictions = glob("*/2024-10-18/preds*")
for file in predictions

=#
#=
For Kahurangi Data
a=glob("*/*/*.csv")
b=glob("*/*/*/*.csv")
c=glob("*/*/*/*/*.csv")
list=[a ; b ; c]
# to delete empty preds.csv files
for file in list
    size = stat(file).size
    if size < 10
        println("Deleting $file - $size")
        rm(file)
    end
end
## change date on preds below (no longer required at 25/10/24)
function make_clips_kahurangi(preds_path::String, label::Int = 1)
    # Assumes function run from Kahurangi Data
    #pth = replace(preds_path, "preds-2024-10-21.csv" => "")
    pth = split(preds_path, "/") |> x -> joinpath(x[1:end-1]) |> x -> x * "/"
    function assert_detections_present_(df::DataFrame, label::Int, preds_path)::DataFrame
        label in levels(df.label) ? (return df) :
        @error "No detections for label = $label at $preds_path"
    end
    # Load and group data frame by file
    gdf =
        #! format: off
        DataFrame(CSV.File(preds_path)) |>
        x -> Skraak.assert_not_empty(x, preds_path) |>
        x -> Skraak.rename_column!(x, "1.0", "label") |> #can remove now, needs to be label
        x -> assert_detections_present_(x, label, preds_path) |>
        x -> Skraak.filter_positives!(x, label) |>
        Skraak.group_by_file!
        #! format: on
    # Make clip and spectrogram
    for (k, v) in pairs(gdf)
        #file_name = chop(v.file[1], head = 2, tail = 4)
        file_name, extension = Skraak.path_to_file_string(v.file[1])
        start_times = v[!, :start_time] |> sort
        detections = Skraak.cluster_detections(start_times)
        isempty(detections) && continue
        signal, freq = Skraak.wavread("$pth$(file_name).$(extension)")
        if freq > 16000
            signal, freq = Skraak.resample_to_8000hz(signal, freq)
        end
        freq = freq |> Float32
        length_signal = length(signal)
        for detection in detections
            st, en = Skraak.calculate_clip_start_end(detection, freq, length_signal)
            p=chop(pth, tail=1) |> x -> replace(x, "/" => "---") #delete trailing / then replace / with ---
            #name = "$p-$file_name-$(Int(floor(st/freq)))-$(Int(ceil(en/freq)))"
            name = "$file_name-$(Int(floor(st/freq)))-$(Int(ceil(en/freq)))" #leave off path, not necesaray
            ##name = "$location-$h-$trip_date-$file_name-$(Int(floor(st/freq)))-$(Int(ceil(en/freq)))"
            f = "Clips_$(today())"
            mkpath(f)
            outfile = "$f/$name"
            sample = signal[Int(st):Int(en)]
            Skraak.wavwrite(sample, "$outfile.wav", Fs = Int(freq))
            image = Skraak.get_image_from_sample(sample, freq)
            PNGFiles.save("$outfile.png", image)
        end
        print(".")
    end
    print(".")
end
using Glob, Skraak, CSV, DataFrames, Dates, PNGFiles
a=glob("*/*/preds-2024-08-29.csv")
b=glob("*/*/*/preds-2024-08-29.csv")
c=glob("*/*/*/*/preds-2024-08-29.csv")
predictions = [a ; b ; c]
## or
predictions = glob("*/*/preds-2024-10-21.csv")
for file in predictions
try
make_clips_kahurangi(file)
catch x
println(x)
end
end