Next month prediction

[tqAkshay95]

Hello,
I am working on creating a workflow in KNIME Analytics Platform to predict accuracy for the next month from a given dataset using the Predictor node. However, I am uncertain about the expected results. Please suggest a workflow design with configurations.

[rfeigel]

I assume you want to predict “achieved” for the next month. How large is your data set? This could be a non-trivial problem if there is seasonality in your data. Finally, if you can upload your data it would be a lot easier to get help.

[mlauber71]

@tqAkshay95 you might want to start by reading these two threads from the forum and think about what variable would be your traget and which variables would actually predict the outcome (target) and if you have any chance of deriving one from the other.

Prediction of cost based on the historical data

@singhsukh12 you will have to decide if this is a problem that indeed can be formulated as a regression task, this will depend on the relationship between your variables and he target. A collection of regression algorithms can be found here Otherwise this might be a time series problem if the costs are seasonal.

Predicting customer sales

You could check out the collection of articles about regression models (not all of them are about time series): The collection about time series might be especially interesting: For future predictions you would typically have to provide a time series model that would accept future month and day dates (derived from historical data) or you might think about giving you historical data something like index numbers running from 1 … 1000 (or whatever your time period is) and then assign that to …

Here you have some small examples from the forum for example about predicting unemployment numbers for the UK; with all the proper warnings why this might be a futile endeavour:

Sales Forecast with recursive loop

@Patrick92 welcome to the KNIME forum Like @Snowy said: it would be helpful if you could give us some more details. Then. We had a somewhat similar discussion about the prediction of numeric targets (unemployment in the UK) here: This resulted in a collection of workflows (please consider the whole workflow group): Or this: Then there is this collection about machine learning methods. In your case a regression model might be the right one. Or you might have to use something to predict…

If you decide to go for a time series approach here is a collection of resources how to to that:

Getting started with time series

@polina_l there is a dedicated book as well as examples on the KNIME Hub you could begin with Codeless Time Series Analysis with KNIME KNIME Hub Space - Codeless Time Series Analysis with KNIME (the examples from the book) Tutorial: Introduction to Time Series Analysis Time Series Analysis with KNIME — an introduction Time Series Analysis Workshop [Time Series Analysis Workshop] Time Series Analysis with the KNIME Analytics Platform [Time Series Analysis with the …

If you go full regression (numeric target) here is a collection of methods and metrics in KNIME:

Score Kaggle House Prices: Advanced Regression Techniques - prepare data...

But maybe you first want to tell us more about the data and your task and if possible maybe upload it (without spelling any secrets of course).

[tqAkshay95]

Thank you. Please find the attached Excel file with data. I’ve been trying various approaches, but I’m confused about the actual output. Could you please help me correct this?

marData.xlsx (3.8 KB)

[mlauber71]

@tqAkshay95 this will not be enough. This data will not enable you to have a decent model I assume. Maybe you can take another look at what I wrote and provide additional information.

[rfeigel]

I agree with @mlauber71

[Daniel_Weikert]

Single Exponential Smoothing (SES). Results would probably wide of the mark without Trends or Seasonal Factors included. Maybe just assume prior year value?
Just trying to be pragmatic here!
@mlauber71 probably said everything
br

[tqAkshay95]

@mlauber71 What type of data is needed? Can you provide a sample?

[tqAkshay95]

Hello @takbb, @mlauber71 , @rfeigel, @Daniel_Weikert can we use this type of data to design a predictor module? Please check the following Excel file and provide a solution.

marketing_data.xlsx (14.0 KB)

[mlauber71]

@tqAkshay95 I think you will have to tell us more about the data, the task and what it is you want to do. How many entries are there, what is the target variable, what does it tell us. Also you might benefit from taking a few courses on data science and model development in general (Learn data science with resources for all learner types | KNIME).

Also: the file seems to contain personal information so I took the liberty of informing the moderators.