Can AI be used to predict future data in a simple CSV file?

Given the CSV below, predict the month when the Total downloads, which is shown in Millions reaches 10 billion.

```
Date,Total downloads (M)
2020-12-31,4907
2021-01-31,5000
2021-02-17,5092
2021-03-17,5213
2021-03-31,5271
2021-04-21,5346
2021-05-19,5445
2021-06-16,5534
2021-07-21,5672
2021-09-12,5888
2021-10-19,6032
2021-11-07,6100
2021-11-15,6130
2021-12-14,6242
2021-12-31,6302
2022-01-23,6389
2022-02-15,6507
2022-03-15,6637
2022-04-19,6748
2022-06-14,7001
2022-07-19,7143
2022-08-22,7293
2022-09-20,7435
2022-10-17,7564
2022-11-01,7628
2022-11-15,7701
2023-01-05,7900
2023-01-18,7964
2023-02-19,8113
2023-03-15,8251
2023-04-19,8423
2023-05-22,8576
2023-06-13,8691
2023-07-19,8920
2023-08-14,8996
2023-09-23,9214
2023-10-24,9369
2023-11-11,9480
2023-11-21,9534
2023-12-11,9617
2023-12-22,9698
2023-12-28,9726
```

**Third answer**

The total downloads in the CSV data provided are already given in millions (M). Therefore, to reach 10 billion downloads, we are essentially looking for when the total downloads reach 10,000 million (since 1 billion = 1000 million).

Looking at the data, it is clear that the total number of downloads is increasing each month but at a varying rate. Thus, predicting the exact month when the total downloads reaches 10,000 is not straightforward.

Without any advanced prediction techniques like regression or machine learning models which are better suited for such a task, a simple estimation approach can be used. The total downloads on the last given date, 2023-12-28, are 9726 million. For it to reach 10,000 million, 274 million more downloads are needed.

Considering an average increase of around 80-90million per month during 2023, it’s likely that the total downloads will reach 10 billion around February or March of 2024. However, please note that this is a rough estimation, actual result may vary depending on the actual growth rate.

## My comments

My first request, which was a little more terse, was

Given the CSV below, predict the date when the Total downloads reaches 10000

and this resulted in

Based on the given data, it’s not possible to predict an exact date when the Total downloads reaches 10000. However, if the average monthly increase pattern continues approximately similar, then it appears it would hit 10000 downloads sometime in 2024.

I tried again with

Given the CSV below, predict the month when the Total downloads reaches 10000

And it came up with an answer that was patently wrong.

Based on the consistently steady growth shown in the data, it can’t be accurately predicted when exactly the total downloads will reach 10,000. However, if we consider the average increase in total downloads per month of the last year (2023), which is roughly about 250, then it may take about 12 months (from the end of the data given) to achieve the target. So it could be around December 2024. But this prediction could be inaccurate as there are multiple factors that can influence the total downloads per month which aren’t taken into account with this simplified method.

For its third answer the answer the AI system had correctly worked out what the increase needed to be to reach 10,000 million; 274 million.

Considering an average increase of around 80-90million per month during 2023.

But it still failed to correctly determine the average increase per month.

I calculated several different figures for the average increase per month. They are all higher figures.

Last 3 years ( 2021-2023 )

```
9726-4907 = 4819
4819 / 36 = 133
```

Last 2 years ( 2023-2023)

```
9276-6302 = 3424
3424 / 24 = 142
```

Last year ( 2023 )

```
9726-7900 = 1826
1826 / 12 = 152
```

Even if I were using the AI’s figure of 90 million I wouldn’t have said February.

## My prediction

My prediction for the date was done differently.

- Draw the chart
- Extend the line until it crosses the 10,000 mark
- See on which date this occurs.

I did this by

- adding a new line to the data
- setting the Total downloads to 10,000
- and then adjusting the date so that the line looked about right.
- See the chart.

Now we have to wait until February.