Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir

HI Asgeir, 

1. Unfortunately, for the forecast, the sandbox dimension cannot be either in the context or the bench. MDX wise there is no difference between having those dimension on bench or context. If you pick a sandbox in the view then the forecast will be written to that sandbox. However I do see why a user may be confused since the message does not mention Bench
2. If the forecast  (dimension) columns are changed it should clear the values, the redundant message seem like a bug.
3. From what I can see in the chart, the forecast is performing well, for the algorithm to work and detect the seasonality automatically, at least 2 seasons are required (otherwise there is no repeatable pattern), what I can see on the chart is  about 1 season and a half, in some cases inputting the seasonality manually will help, but the best option is to add more historical data.

Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir

HI Asgeir, 

1. Unfortunately, for the forecast, the sandbox dimension cannot be either in the context or the bench. MDX wise there is no difference between having those dimension on bench or context. If you pick a sandbox in the view then the forecast will be written to that sandbox. However I do see why a user may be confused since the message does not mention Bench
2. If the forecast  (dimension) columns are changed it should clear the values, the redundant message seem like a bug.
3. From what I can see in the chart, the forecast is performing well, for the algorithm to work and detect the seasonality automatically, at least 2 seasons are required (otherwise there is no repeatable pattern), what I can see on the chart is  about 1 season and a half, in some cases inputting the seasonality manually will help, but the best option is to add more historical data.

Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir

Here is something to think about for designers/developers of PAW that comes from my old software engineer heart 

When selecting an option like "Seasonality" it is confusing for the user if you change the status icon and also the label for that option. 

The image below indicates that the seasonality is not manual

HI Asgeir, 

1. Unfortunately, for the forecast, the sandbox dimension cannot be either in the context or the bench. MDX wise there is no difference between having those dimension on bench or context. If you pick a sandbox in the view then the forecast will be written to that sandbox. However I do see why a user may be confused since the message does not mention Bench
2. If the forecast  (dimension) columns are changed it should clear the values, the redundant message seem like a bug.
3. From what I can see in the chart, the forecast is performing well, for the algorithm to work and detect the seasonality automatically, at least 2 seasons are required (otherwise there is no repeatable pattern), what I can see on the chart is  about 1 season and a half, in some cases inputting the seasonality manually will help, but the best option is to add more historical data.

Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir

Hi Asgeir, 

Few ideas, 

1. Move the sandbox dimension to the rows
2. Makes sense,
3. Seasonality is not auto-detected, because there is no repeatable pattern in the data (as I said, you can see about 1 season and a half in the chart), for the algorithm to pick a value or trend just as in  "the previous month last year", it will have to get the seasonality, so you need a repeated pattern, in the chart there iso.  the Unable to get the preview could be due to insufficient points for the 12 time periods input (but deserves some investigation). More data should give you better results (you mentioned 2 years of data, but at least in the chart only 1 year and 8 months are shown as historical data), In short this pattern must be present at least twice :
   
   

Here is something to think about for designers/developers of PAW that comes from my old software engineer heart 

When selecting an option like "Seasonality" it is confusing for the user if you change the status icon and also the label for that option. 

The image below indicates that the seasonality is not manual

HI Asgeir, 

1. Unfortunately, for the forecast, the sandbox dimension cannot be either in the context or the bench. MDX wise there is no difference between having those dimension on bench or context. If you pick a sandbox in the view then the forecast will be written to that sandbox. However I do see why a user may be confused since the message does not mention Bench
2. If the forecast  (dimension) columns are changed it should clear the values, the redundant message seem like a bug.
3. From what I can see in the chart, the forecast is performing well, for the algorithm to work and detect the seasonality automatically, at least 2 seasons are required (otherwise there is no repeatable pattern), what I can see on the chart is  about 1 season and a half, in some cases inputting the seasonality manually will help, but the best option is to add more historical data.

Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir

Thank you, Julian

It would be useful if I could create a prediction from one year's pattern (as you highlight in red) and the trend in the past few months. 

The patterns from the past are sometimes irrelevant and if you think of the covid periods then everything was skewed. 

Hi Asgeir, 

Few ideas, 

1. Move the sandbox dimension to the rows
2. Makes sense,
3. Seasonality is not auto-detected, because there is no repeatable pattern in the data (as I said, you can see about 1 season and a half in the chart), for the algorithm to pick a value or trend just as in  "the previous month last year", it will have to get the seasonality, so you need a repeated pattern, in the chart there iso.  the Unable to get the preview could be due to insufficient points for the 12 time periods input (but deserves some investigation). More data should give you better results (you mentioned 2 years of data, but at least in the chart only 1 year and 8 months are shown as historical data), In short this pattern must be present at least twice :
   
   

Here is something to think about for designers/developers of PAW that comes from my old software engineer heart 

When selecting an option like "Seasonality" it is confusing for the user if you change the status icon and also the label for that option. 

The image below indicates that the seasonality is not manual

HI Asgeir, 

1. Unfortunately, for the forecast, the sandbox dimension cannot be either in the context or the bench. MDX wise there is no difference between having those dimension on bench or context. If you pick a sandbox in the view then the forecast will be written to that sandbox. However I do see why a user may be confused since the message does not mention Bench
2. If the forecast  (dimension) columns are changed it should clear the values, the redundant message seem like a bug.
3. From what I can see in the chart, the forecast is performing well, for the algorithm to work and detect the seasonality automatically, at least 2 seasons are required (otherwise there is no repeatable pattern), what I can see on the chart is  about 1 season and a half, in some cases inputting the seasonality manually will help, but the best option is to add more historical data.

Hi all

I guess I am just reporting some bugs in the PAW forecasting option 

1. Not able to preview due to "Sandboxing" error
2. Error message each time I change some dimension, Asks me to select the forecasting periods again even though they are visibly still selected. 
3. When the preview works based on periodical data, the result is useless. 

---

1. Not able to preview due to "Sandboxing" error 
The Base sandbox dimension is in the "Bench", not in the "Context Bar" but I still get this error

2. Error message each time I change some dimension, 

Asks me to select the forecasting periods again even though they are visibly still selected. 

The start date is selected but I always get this message

3. When the preview works based on periodical data, the result is useless. 

Based on these periodical values, this forecasting result makes no sense, ..at least will not help

Thanks, Asgeir