Statistics about the amount of saved food

That is probably again the case of foodsaving groups working differently. When people save the food at a store here they will always see the total amount of food saved and can guess on that, even if they don’t go all the way to the sharing spot.

There is definitely not a one-size-fits-all solution for this feature, so I think we need to continue to investigate more the different realities of the groups using Karrot before changing the feature.

Not really. Sometimes there is so much food that it’s not really possible to assess the total. And we rather not guess the amount of saved food but either weigh it or estimate it, for example, by knowing the average weight of the item and multiplying it by its count. We don’t do it during the pickup though which has to be done quickly but later when we take food home to pack it into portions ready to be left at a sharing spot. That’s why each Foodsaver cannot take responsibility for the total amount because they don’t know total amount. We want to keep stats reliable and we don’t want our users to just guess what the total amount should be.

In your case, when each Foodsaver is able to assess total amount, why couldn’t they assess the amount they are carrying? Total will be a sum of individual assessments. It’s not that different than an average of individual assessments.

I assume you’re speaking for your group. In our case we actually do an assessment even if it’s hundreds of kilos, by multiplying the number of banana-boxes (the standard we use in transporting in many cases) by 10 kg, or just by a simple visual assessment. That is obviously inaccurate and a gross estimation, but it’s not our concern (and I believe most of the groups) to spend time weighing things properly.

Usually the people who do use the feature, which are a minority here, are the ones with a bit more experience and feeling more confident about guessing the total amount.

Because many times they do the collection and transportation of the food together and it does not make sense to think only of the amount they are individually carrying. We could say then that people would input the amount of food they got to themselves, that is, what they brought home with them. That’s another measurement. The feature could be changed and used to estimate that, but then it should be clear and then our group (and maybe other groups that leave considerable amounts at sharing spots) would have to decide on one person who reports their personal share plus the amount of food left in sharing spots.

If you are not really interested in having reliable stats, why does it matter to you how this feature works? :slight_smile: In our group 99% of pickups have quite detailed feedback with quite detailed weights. That’s why I’m insisting on not spoiling our evaluations by averaging the amounts.

This is hardly the purpose of this feature. We want to keep record of saved food, not food taken home.

Maybe one person can give feedback with total amount then? If not many users in your group use this feature, it shouldn’t be a big deal, right?

to mzpawlowshi: I share your opinion completely!

E.g. for us (EfA - in Germany) the feetback feature, with detaild and realistic amounts for each shop and each foodsaver, is crucial and we use it all the time.


Especially if so many foodsaving groups are working in a different manner
an algorithm to built an everage or what ever will confuse people, would make the system less detailed and correct but more complicated.

In our group we often pickup the staff with several persons and in most cases, about 90% will be carried e.g. to varous Kindergarden and so on. Only about 10% is for private use.

For example 3 foodaver are picking up the total amount of about 50 kg net weight of goods (excluding the rubish) in one market.
1 foodsaver is carrying 45 kg fruits and vegetable for Kindergarden No1,
1 is taking 5 kg chocolate for Kindergarden No2, the other foodsaver helped but carried 0 kg.
We would prefere, if everyone can state the real amount and may give comments.
Statistic should show 50 kg as tolal amount, and nothing else (not a total of 25kg and not a total of 15kg fruit-vege-late).

1 Like

Oh, what a heated discussion! I’d like to take a step back and would be interested why the pickup weight is useful and/or important to you. That would help me a lot to implement good statistics :smiley:

Why the pickup weight is important for us:

Pickup weights are quite importtant for us,
because we report exactly

  • how much and what we received from each market in total,
  • how much and what everyone of us has taken,
  • and how much and what we have given to each institution.

Since we are a registered non-profit organisation we are obliged to report that we spend most of the stuff we received.
A well working statistic-tool in karrot is the best instrument to proof our obligations.

I’m definitely not making a big deal of how the feature should be and how statistics should be collected, but I’m making a big deal of the following points:

  • Not every group works the same and we should take that into account when designing and changing this feature. Maybe we should ask for more input from other groups? If most groups would like amounts to be set individually then I see no problem; But, more importantly:
  • If you don’t get the semantics correctly, there’s no point in having this feature if people are not understanding and using it accordingly. It became clear that people were misled by some bad translation, and if the feature changes it should be clear on what exactly it is asking for when you set an amount and how it calculates. If changed, it should state “How much food did you individually carry home or to a sharing spot?”. That should do the work fine in groups where you require foodsavers to always give feedback (since one feedback missing might already affect considerably the outcome) and not so much on groups that like to use this feature just to have a rough estimate.

There are no explanations for the user in karrots.
The use is intuitive.
Therefore the rule for every feature should be “what you see is what you get”.

If every person of the pickup-team is asked for feedback by the system and all feedbacks of the members of a pickup-team are shown, the user must assume, that his feedback will be used “pure”, without any hidden or additional algorithm.

I guess the phrase @bruno is refering to is this one:
image
The German version is equally clear stating ‘Menge des insgesamt geretteten Essens’, whereas the Polish version indicates that it’s just about the amount that you yourself saved. Just for context…

(And I’d like to point out that intuition works differently from one person to the other. The use of this feature is the perfect example to illustrate that imho… :wink: )

1 Like

I’m revisiting this topic just now and I thought a feedback export feature might be useful to you. You would be able to download a CSV file that you can import in LibreOffice or Excel. This would be per place or for the whole group.

You could use this file to calculate all kinds of statistics, e.g. weekly weight, average weight per pickup or weight per user.

Here’s how an example export looks like:

Do you think that would be useful to you? Especially @mzpawlowski @karolina @sylviaschneider

2 Likes

It’s great! Very helpful indeed!

Would it be also possible to extract CSV files with data other than feedbacks? E.g. history of pickups in a place or a history of user’s actions, etc.?

1 Like

Sure, we just need to figure out what columns the file should contain and which types to include. History data is a bit hard to work with because every type has its own payload/context.

1 Like

Great! I think the CSV file will do.
Best regards
Sylvia

Hey you statistics people! I released the first version of the CSV feedback export. It’s just a URL for now; there is no clickable link in Karrot (yet).

You can export all feedback with this:
https://karrot.world/api/feedback/export/

If you want it just for one group, pass the group id like this (16 is the playground group):
https://karrot.world/api/feedback/export/?group=16

And the same for just one place:
https://karrot.world/api/feedback/export/?place=72

You can find the group and place ids in your browser, e.g. https://karrot.world/#/group/16/place/72/pickups means group 16 with place 72.

Some more filtering examples:

1 Like

I’ve extracted a file for Foodsharing Warsaw (group=39) and successfully imported it into Excel which is great :slight_smile: The only comment which I have is that a file doesn’t have CSV extension after downloading it. It’s an uknown format but it still works.

Great job!

Glad that it works!

That depends on the browser. I noticed that it has a CSV ending in Firefox, but not in Chrome.

I find extract with feedbacks very useful but I don’t hide I’m really waiting for the extracts with history. Have you already had a chance to work on this? Maybe you have something which works to some extent but could already be used/tested?

Hey @mzpawlowski - I have a code branch that implements the basics, however I still need to rewrite old history entries to contain the same data as new ones and implement a fallback where this isn’t possible.

Chances are good that I will get around to do it in the next weeks, now that I’m almost done with upgrading Quasar, our user interface library. That took more than a month already and I’m looking forward to code something more fulfilling :wink: But it made pretty clear to me how much the Karrot grew in the last years…

1 Like

Coming back to this thread - I implemented https://github.com/yunity/karrot-frontend/issues/1455, which means these things will change once I do the release (probably in the next days):

  • multiple feedback entries per pickup will be summed, not averaged
  • calculations for historical feedback will not change
  • a new clarification text on the feedback form: “For place statistics, we will add together the amount of all feedback.”

Does this sound alright?

In case you already want to translate the new text before the release, visit https://www.transifex.com/yunity-1/karrot/translate/

1 Like