Dear All,
Is there a possibility to predict for next two years-2017 and 2018 with the following data? For Revenue, COGS and Gross profit?
I want to use ARIMA.
Please find the data below:
2013 | 2014 | 2015 | 2016 | |
Revenue | 10,000 | 15,000 | 20,000 | 25,000 |
COGs | 5000 | 10,000 | 15,000 | 20,000 |
Gross Profit | 5,000 | 5,000 | 5,000 | 5,000 |
Please advise.
Thank you in advance.
Farhana
Solved! Go to Solution.
Hi @farhana_91
As with most of the predictive tools here, they're very easy to get working, you just need to be wary of the: way your data looks; ways in which you're configuring the tools; you're interpreting the results.
There's not a huge amount of data here (do you have more?), and in my experience these work better when looking at a week/month/quarter level, but I've included a short example against your data (with a few years already added) attached here. The key is to ensure your years are in rows, and you tell the model that you're working in years:
For further info, I'd recommend going into your Designer and clicking Help -> Sample Workflows -> Predictive Tool Samples -> Predictive analytics -> 15 Time Series which has some great examples of all those tools
Hope that helps and gives you a start point at least!
Andy
Yes you can do ARIMA all you need is 2 records. Also, there is no point in my opinion of forecasting COGs or Profit because they are directly correlated to Revenue. Just Revenue would suffice and then proportionate the expenses based on history from that predicted value.
Thanks Andy! This works for me as of now!
Agreed! Thanks Ryan!
Great, pleased it worked!
Hi Andy,
Found this post today, it is a very helpful solution. Got a quick question, for the original data set the COGS and GrossProfit were not forecasted , as we think they are related to the revenue. What if now these three dimensions change to different product names (150 products), and I have to forecast the sales for these 150 products. Which tool can I use? The solution you provided previously could only forecast one product a time. I tried the TS Factory won't work, because I only have 24 months data for each product.
Look forward to your reply 🙂
W.
Hi @WCH ,
I thought to chime in and answer to your question.
If you would like to forecast multiple product at once, you can use TS Model Factory Building Block to estimate time series forecasting models for multiple groups at once. To create forecasts you can use TS Forecast Factory Building Block that provides forecasts from groups of either ARIMA or ETS models for a user-specified number of future periods.
You can go here TS-Factory Sample to check out workflow samples created by Alteryx Product team.
I hope this helps.
Chi
Hi Chin,
Thanks for you quick reply. I actually tried the TS model Factory, but failed. 😞
Maybe I didn't set up the configuration, as I don't really know how to choose it, such as the target field, and the grouping field.
Also , it keeps telling me an error: insufficient number of records was provided for the group.
I attached the sample file for your reference, actually my data is quite simple, it just doesn't work 😞
Hi @WCH
Please try to put your months in rows instead of columns. You can use Transpose building block for this. Hope this helps.