Software for cellar inventory reccomendation?

[Scott_Brunson]

The only problem with CellarTracker is the pending delivery functionality. It somehow shows a lot of bottles automagically because I told my wife I wasn’t buying much wine anymore…so the program must have a bug.

I had the same bug, I changed my password, presto, problem solved . . .


Until the boxes showed up

Offsite storage

[Eric_LeVine]

Sorry, I haven’t been checking in here too often lately. Thanks for all the kind words about CellarTracker. I have been at it for 17.5 years and been committed to not doing creepy shit, letting people control their own data, having a flexible payment model etc. I think our values are pretty clear in this regard.

Anyway, Michael or others, if you have existing data and want a faster way to load it up, we have a nice bulk import tool. Just shoot us a note at support@cellartracker.com. We are a small but slowly growing team of 4, just chipping away steadily.

[Shan_A]

been committed to not doing creepy shit

Thank you.

[Alex_N]

Going off on a tangent…

Sure, most cloud services are robust from a technical perspective (or at least that’s the intention; implementation doesn’t always follow suit). The risk is that in the blink of an eye, you may find yourself unable to use them and unable to access your data. It could be the result of hacking, buyout, a new business model, government action, incompetence, whatever. There’s a multitude of risks that are almost impossible to quantify.

Granted, this matters more to somebody running a business than your average wine berserker, but it astonishes me how quickly people and organizations have given up control of their data.

Cellar Tracker has an export to .csv/Excel feature. So long as you take a copy of your data semi-often I see no reason to be concerned.

[jrozes]

Sorry, I haven’t been checking in here too often lately. Thanks for all the kind words about CellarTracker. I have been at it for 17.5 years and been committed to not doing creepy shit, letting people control their own data, having a flexible payment model etc. I think our values are pretty clear in this regard.

Anyway, Michael or others, if you have existing data and want a faster way to load it up, we have a nice bulk import tool. Just shoot us a note at support@cellartracker.com. We are a small but slowly growing team of 4, just chipping away steadily.

Yes, I should have mentioned that I’ve always found CT (Eric, specifically) to be far more transparent than most and I’ve been a happy (paying) customer for six years.

[Sarah_Kirschbaum]

Even if you don’t use CT for your own cellar management (and I think everyone should), it’s one of the largest and best repositories of information out there when it comes to certain kinds of details about producers and wines. Many questions I have had have been faster to answer via CT than Google.

[HenryB]

Sorry, I haven’t been checking in here too often lately. Thanks for all the kind words about CellarTracker. I have been at it for 17.5 years and been committed to not doing creepy shit, letting people control their own data, having a flexible payment model etc. I think our values are pretty clear in this regard.

Anyway, Michael or others, if you have existing data and want a faster way to load it up, we have a nice bulk import tool. Just shoot us a note at support@cellartracker.com. We are a small but slowly growing team of 4, just chipping away steadily.

Love the service, and more than happily a paying customer.


One question though - I’ve started recently looking at Machine Learning or general analytics in the wine space. It’s fairly trivial to generate a ‘critic matcher’ - e.g. of these critics, scores wine most similarly to you. I’d love to generate a wine recommender, but it would take something more akin to the database that Cellartracker had to be able to do that. Is this an area of interest at all for you guys?


(Critic matcher was done for personal interest, I wanted to look at some of the basic statistics, e.g. consistency of scores across critics, the linear model between, say, suckling vs parker, etc. I had considered making it a free public service to help people find their favourite critics, but its a clunky interface unless you can get something like Cellartracker integration tbh!)

[Michael_D]

Sorry, I haven’t been checking in here too often lately. Thanks for all the kind words about CellarTracker. I have been at it for 17.5 years and been committed to not doing creepy shit, letting people control their own data, having a flexible payment model etc. I think our values are pretty clear in this regard.

Anyway, Michael or others, if you have existing data and want a faster way to load it up, we have a nice bulk import tool. Just shoot us a note at support@cellartracker.com. We are a small but slowly growing team of 4, just chipping away steadily.

Thank you. I just registered and sent you an Excel file to upload. I look forward to giving this a whirl.

[John_Glas]

As mentioned numerous times cellar tracker. I don’t how many times I am at a wine shop looking to buy something of interest and plug it into cellar tracker for info. It is always nice to see what I scored the wine and makes for easy purchasing decisions. Not to mention Eric is a great guy and has a great product.

[Alex_Valdes]

Sorry, I haven’t been checking in here too often lately. Thanks for all the kind words about CellarTracker. I have been at it for 17.5 years and been committed to not doing creepy shit, letting people control their own data, having a flexible payment model etc. I think our values are pretty clear in this regard.

Anyway, Michael or others, if you have existing data and want a faster way to load it up, we have a nice bulk import tool. Just shoot us a note at support@cellartracker.com. We are a small but slowly growing team of 4, just chipping away steadily.

Love the service, and more than happily a paying customer.


One question though - I’ve started recently looking at Machine Learning or general analytics in the wine space. It’s fairly trivial to generate a ‘critic matcher’ - e.g. of these critics, scores wine most similarly to you. I’d love to generate a wine recommender, but it would take something more akin to the database that Cellartracker had to be able to do that. Is this an area of interest at all for you guys?


(Critic matcher was done for personal interest, I wanted to look at some of the basic statistics, e.g. consistency of scores across critics, the linear model between, say, suckling vs parker, etc. I had considered making it a free public service to help people find their favourite critics, but its a clunky interface unless you can get something like Cellartracker integration tbh!)

I’d love to see the critic to critic analysis. Is there enough variance to say anything definitive? Also, are you correlating the rating levels or deviation from some baseline? Correlating levels is not likely to give you the right answer.

[Freek_S_u_i_j_v_e_r]

CellarTracker is OK. It’s what I use while waiting for MyWines to become available. That is going to be awesome!

[HenryB]

It’s a linear model between critic pairs - I fnid an R^2 of 0.441 (the maximum in the data set) between Parker and Martin, Suckling drops to 0.385, Suckling v Parker is 0.437. What is quite interesting, as you might expect, are the best fits. Suckling = 0.437*parker+31.4 - which is no surprise, there’s general consensus that Suckling scores high.

I didnt take it into too much depth, was focused on building a lightweight MVP only, what really interested me to try to get working well was the critic-recommender, e.g. your scores match XXXXX most closely, so you might want to consider following them.



I tried to build a linear model to infer Jancis’ scores on a 100 point scale, but unfortunately the performance was pretty poor - albeit its known she scores quite differently, to say the least, compared to other critic. Interesting to see how few critics give 100 points out to wines that other critics have given 100 points out (in the data set I used, at least, which is typically very new wines rather than older ones)

[Alex_Valdes]

It’s a linear model between critic pairs - I fnid an R^2 of 0.441 (the maximum in the data set) between Parker and Martin, Suckling drops to 0.385, Suckling v Parker is 0.437. What is quite interesting, as you might expect, are the best fits. Suckling = 0.437*parker+31.4 - which is no surprise, there’s general consensus that Suckling scores high.

I didnt take it into too much depth, was focused on building a lightweight MVP only, what really interested me to try to get working well was the critic-recommender, e.g. your scores match XXXXX most closely, so you might want to consider following them.



I tried to build a linear model to infer Jancis’ scores on a 100 point scale, but unfortunately the performance was pretty poor - albeit its known she scores quite differently, to say the least, compared to other critic. Interesting to see how few critics give 100 points out to wines that other critics have given 100 points out (in the data set I used, at least, which is typically very new wines rather than older ones)

Cool. Would be great to have this on GitHub so others could try to derive insight. Good Kaggle competition as well

[HenryB]

I was tempted to post it, but ahem was a little naughty in sourcing the data.

I’m still keen to get it hosted as a website where people can input either a link to their CT account or a handful of scores of their own, and match to a critic. I’m no software engineer, though, so need help on that piece if you’re interested

[Sean_S]

As mentioned numerous times cellar tracker. I don’t how many times I am at a wine shop looking to buy something of interest and plug it into cellar tracker for info. It is always nice to see what I scored the wine and makes for easy purchasing decisions. Not to mention Eric is a great guy and has a great product.

+100

As so many have said in other threads. Its the one wine tool I literally use everyday.

Sean

[John_Glas]

+100

As so many have said in other threads. Its the one wine tool I literally use everyday.

Sean

Agree. Might be on it at least 320 days plus a year.

[Andy_Sc]

I tried to build a linear model to infer Jancis’ scores on a 100 point scale, but unfortunately the performance was pretty poor - albeit its known she scores quite differently, to say the least, compared to other critic. Interesting to see how few critics give 100 points out to wines that other critics have given 100 points out (in the data set I used, at least, which is typically very new wines rather than older ones)

We made a series of tastings where all participants scored the wines blind. Roughly half of the group used the 20 points scale, the other half the 100 points scale. We always translated the scores according to the “official” 20 to 100 points conversion table but that never really worked out as the average score of the 20 points scale crowd was always considerably lower.

Then, one day, I’ve met French critic JM Quarin and he explained to me that his formula to translate 20 points to the 100 points scale is fairly simple. 20=100, 19.5=99, 19=98, 18.5=97… (vs 20=100, 19.75=99, 19.5=98, …). When I told that idead to my group, the 20 points crowd rebuffded Quarin’s idea as they said that their scores would be too high.

Funny thing is, when applying the Quarin conversion onto our various tastings, it showed that the two groups no longer where far appart in terms of average score, but very close together. Indeed the Quarin method proofed to be the perfect conversion, making scores perfectly comparable. Maybe you should try this too.

[Otto_Forsberg]

(vs 20=100, 19.75=99, 19.5=98, …)

Where do you get the logic for this one?

To me the most logical would be a direct conversion:

If the other system has a maximum of 100 and a theoretical minimum of 50, while other has max 20 and min 0, it means the other system is 2.5x larger than the other, plus has 50 extra points for kicks-and-giggles, the most logical system would be to

• first deduct those extra 50 points and then divide by 2.5 in 100-point system
• first multiply by 2.5 and then add additional 50 points in 20-point system.

The result doesn’t correspond to either of your systems entirely, but the results are closer to the one suggested to you than to the one I quoted above.

For example
95 points minus 50, then divided by 2.5 = 18 pts.
16.5 points multiplied by 2.5, then +50 = 91(.25) pts.

[HenryB]

I tried to build a linear model to infer Jancis’ scores on a 100 point scale, but unfortunately the performance was pretty poor - albeit its known she scores quite differently, to say the least, compared to other critic. Interesting to see how few critics give 100 points out to wines that other critics have given 100 points out (in the data set I used, at least, which is typically very new wines rather than older ones)

We made a series of tastings where all participants scored the wines blind. Roughly half of the group used the 20 points scale, the other half the 100 points scale. We always translated the scores according to the “official” 20 to 100 points conversion table but that never really worked out as the average score of the 20 points scale crowd was always considerably lower.

Then, one day, I’ve met French critic JM Quarin and he explained to me that his formula to translate 20 points to the 100 points scale is fairly simple. 20=100, 19.5=99, 19=98, 18.5=97… (vs 20=100, 19.75=99, 19.5=98, …). When I told that idead to my group, the 20 points crowd rebuffded Quarin’s idea as they said that their scores would be too high.

Funny thing is, when applying the Quarin conversion onto our various tastings, it showed that the two groups no longer where far appart in terms of average score, but very close together. Indeed the Quarin method proofed to be the perfect conversion, making scores perfectly comparable. Maybe you should try this too.

This was kind of the point of building a linear model to try to accuratelly estimate it - it doesnt rely on any hardwired translation system, but a linear model that is data-defined to estimate a 20-scale translation to 100-scale. It’s foiled because of the inconsistency off scores by different critics for the same wine, of course, but that is no surprise really!

[Andy_Sc]

(vs 20=100, 19.75=99, 19.5=98, …)

Where do you get the logic for this one?

To me the most logical would be a direct conversion:

If the other system has a maximum of 100 and a theoretical minimum of 50, while other has max 20 and min 0, it means the other system is 2.5 bigger than the other, plus has 50 extra points for kicks-and-giggles, the most logical system would be to

• first deduct those extra 50 points and then divide by 2.5 in 100-point system
• first multiply by 2.5 and then add additional 50 points in 20-point system.

The result doesn’t correspond to either of your systems entirely, but the results are closer to the one suggested to you than to the one I quoted above.

For example
95 points minus 50, then divided by 2.5 = 18 pts.
16.5 points multiplied by 2.5, then +50 = 91(.25) pts.

https://www.texaswinejournal.org/20-point-to-100-point-conversion/

How to read Decanter wine scores: the 100 point scale

A Detailed Conversion Table for the 20-Point Scale to 100-Point Scale |...

Lot of people use that conversion, when the most accurate I’ve found is Quarin one. I find your way quite interesting (with the results being closer to the simple Quarin method than to the other method of Decanter and others).