Another rating formula : the energy of the article
-
I like it (at least for high ranked articles). What about low ranking articles with lots of votes? (guess I could pull out the calculator myself but call me lazy ;)) (apologies for reformatting your post - I wanted others to see your idea more clearly) cheers, Chris Maunder
The last Top Ranked Articles normally have few votes (<10), so the "energy" is low. Anyway a "3/5" article needs 50 votes to overcome a "5/5" article with 1 vote.
-
If you compare the "Top Viewed Articles" with the "Top Ranked Articles", you find that the most popular articles have a relatively low rating. This makes me think that the ranking is not a good parameter, especially for the new competitions ;) I suggest this formula: Energy = rating * log10(max(10,#votes)); In this way an article with "4/5" and 50 votes (energy=6.78) is better that an article with "5/5" and 7 votes only (energy=5) The energy for the top viewed articles will be:
#votes rank energyMFC Grid control 333 4.84 12.21
The Code Project Search Bar 63 4.05 7.29
A set of ADO Classes 82 4.80 9.19
Developing a Truly Scalable 87 4.25 8.24
CButtonST v3.5 137 4.64 9.91
Visual Studio .NET Menu 26 4.38 6.19
CXImage 74 4.68 8.75
CResizableDialog 53 4.53 7.81
Cool Owner Drawn Menus 95 4.82 9.53
Layout Manager for Dialogs 94 4.85 9.57The energy for the top ranked articles will be:
#votes rank energyCustom Tab Controls 12 5.00 5.40
Zoom+ 7 5.00 5.00
Get the real XP look 3 5.00 5.00
Drag and Drop Tab Control 2 5.00 5.00
A high performance TCP/IP 1 5.00 5.00
BUG: MFC 7.0's CByteArray 1 5.00 5.00
Drawing Barcodes in Windows 1 5.00 5.00
SIngleton Pattern 1 5.00 5.00
Undocumented Visual C++ 121 4.96 10.33
Achieving PostScript and Wmf 25 4.96 6.93Just realised that you've written
Energy = rating * log10(max(10,#votes))but your calculations are usingEnergy = rating * log10(#votes). I was originally wondering why the votes would be clipped at 10. Your formula doesn't handle the case of an article rated 3 with 100 votes vs. an article rated 5 with 10 votes.#votes rating popularityArticle A 10 5.00 5.00
Article B 100 3.00 6.00What about something like:
pop = (rating-3) * Log10(#votes). Then#votes rating popularity 100 4.50 3.00 20 5.00 2.60 10 5.00 2.00 20 4.50 1.95 5 5.00 1.40 100 3.50 1.00 3 4.50 0.72 5 4.00 0.70 2 5.00 0.60 1 5.00 0.00 100 3.00 0.00 10 2.50 -0.50 100 1.90 -2.65Unfortunately this breaks down if you are close to 3 but have a large number of votes. cheers, Chris Maunder
-
Just realised that you've written
Energy = rating * log10(max(10,#votes))but your calculations are usingEnergy = rating * log10(#votes). I was originally wondering why the votes would be clipped at 10. Your formula doesn't handle the case of an article rated 3 with 100 votes vs. an article rated 5 with 10 votes.#votes rating popularityArticle A 10 5.00 5.00
Article B 100 3.00 6.00What about something like:
pop = (rating-3) * Log10(#votes). Then#votes rating popularity 100 4.50 3.00 20 5.00 2.60 10 5.00 2.00 20 4.50 1.95 5 5.00 1.40 100 3.50 1.00 3 4.50 0.72 5 4.00 0.70 2 5.00 0.60 1 5.00 0.00 100 3.00 0.00 10 2.50 -0.50 100 1.90 -2.65Unfortunately this breaks down if you are close to 3 but have a large number of votes. cheers, Chris Maunder
I made a quick search, and the 99.9% of the articles rated 3 or less have less than 40 votes. There are only FEW exceptions, like your article "Visual Studio.NET screen shots" with 2.93/5 and 67 votes (normal popularity : 5.35), all the other are more or less in the mud. I choose "log10(max(10,#votes))" to avoid a handicap for the new entries and because log10 is <1 for values <10.
-
If you compare the "Top Viewed Articles" with the "Top Ranked Articles", you find that the most popular articles have a relatively low rating. This makes me think that the ranking is not a good parameter, especially for the new competitions ;) I suggest this formula: Energy = rating * log10(max(10,#votes)); In this way an article with "4/5" and 50 votes (energy=6.78) is better that an article with "5/5" and 7 votes only (energy=5) The energy for the top viewed articles will be:
#votes rank energyMFC Grid control 333 4.84 12.21
The Code Project Search Bar 63 4.05 7.29
A set of ADO Classes 82 4.80 9.19
Developing a Truly Scalable 87 4.25 8.24
CButtonST v3.5 137 4.64 9.91
Visual Studio .NET Menu 26 4.38 6.19
CXImage 74 4.68 8.75
CResizableDialog 53 4.53 7.81
Cool Owner Drawn Menus 95 4.82 9.53
Layout Manager for Dialogs 94 4.85 9.57The energy for the top ranked articles will be:
#votes rank energyCustom Tab Controls 12 5.00 5.40
Zoom+ 7 5.00 5.00
Get the real XP look 3 5.00 5.00
Drag and Drop Tab Control 2 5.00 5.00
A high performance TCP/IP 1 5.00 5.00
BUG: MFC 7.0's CByteArray 1 5.00 5.00
Drawing Barcodes in Windows 1 5.00 5.00
SIngleton Pattern 1 5.00 5.00
Undocumented Visual C++ 121 4.96 10.33
Achieving PostScript and Wmf 25 4.96 6.93Seems like a good idea. :-)
"When a friend hurts us, we should write it down in the sand, where the winds of forgiveness get in charge of erasing it away, and when something great happens, we should engrave it in the stone of the memory of the heart, where no wind can erase it" Nish on life [methinks] "It's The Soapbox; topics are optional" Shog 9
-
If you compare the "Top Viewed Articles" with the "Top Ranked Articles", you find that the most popular articles have a relatively low rating. This makes me think that the ranking is not a good parameter, especially for the new competitions ;) I suggest this formula: Energy = rating * log10(max(10,#votes)); In this way an article with "4/5" and 50 votes (energy=6.78) is better that an article with "5/5" and 7 votes only (energy=5) The energy for the top viewed articles will be:
#votes rank energyMFC Grid control 333 4.84 12.21
The Code Project Search Bar 63 4.05 7.29
A set of ADO Classes 82 4.80 9.19
Developing a Truly Scalable 87 4.25 8.24
CButtonST v3.5 137 4.64 9.91
Visual Studio .NET Menu 26 4.38 6.19
CXImage 74 4.68 8.75
CResizableDialog 53 4.53 7.81
Cool Owner Drawn Menus 95 4.82 9.53
Layout Manager for Dialogs 94 4.85 9.57The energy for the top ranked articles will be:
#votes rank energyCustom Tab Controls 12 5.00 5.40
Zoom+ 7 5.00 5.00
Get the real XP look 3 5.00 5.00
Drag and Drop Tab Control 2 5.00 5.00
A high performance TCP/IP 1 5.00 5.00
BUG: MFC 7.0's CByteArray 1 5.00 5.00
Drawing Barcodes in Windows 1 5.00 5.00
SIngleton Pattern 1 5.00 5.00
Undocumented Visual C++ 121 4.96 10.33
Achieving PostScript and Wmf 25 4.96 6.93I think you have the right type of idea, But it might need some tweaking. Remember though these systems will always be controversial. Regardz Colin J Davies
Sonork ID 100.9197:Colin
More about me :-)
-
I think you have the right type of idea, But it might need some tweaking. Remember though these systems will always be controversial. Regardz Colin J Davies
Sonork ID 100.9197:Colin
More about me :-)
the complete formula was:
rating * log10(max(10,#votes))*exp(-(#days since 1st post)/100))+100*(#clicked banners)+10*(#nish's threads);
but maybe it requires too much processing time for the server ;P
-
the complete formula was:
rating * log10(max(10,#votes))*exp(-(#days since 1st post)/100))+100*(#clicked banners)+10*(#nish's threads);
but maybe it requires too much processing time for the server ;P
OMG the Nish factor appears ! It appears to be slightly overwieghted at *10. :-) I can see the issue about processor time on the server, for every page refresh. :-) Regardz Colin J Davies
Sonork ID 100.9197:Colin
More about me :-)
-
If you compare the "Top Viewed Articles" with the "Top Ranked Articles", you find that the most popular articles have a relatively low rating. This makes me think that the ranking is not a good parameter, especially for the new competitions ;) I suggest this formula: Energy = rating * log10(max(10,#votes)); In this way an article with "4/5" and 50 votes (energy=6.78) is better that an article with "5/5" and 7 votes only (energy=5) The energy for the top viewed articles will be:
#votes rank energyMFC Grid control 333 4.84 12.21
The Code Project Search Bar 63 4.05 7.29
A set of ADO Classes 82 4.80 9.19
Developing a Truly Scalable 87 4.25 8.24
CButtonST v3.5 137 4.64 9.91
Visual Studio .NET Menu 26 4.38 6.19
CXImage 74 4.68 8.75
CResizableDialog 53 4.53 7.81
Cool Owner Drawn Menus 95 4.82 9.53
Layout Manager for Dialogs 94 4.85 9.57The energy for the top ranked articles will be:
#votes rank energyCustom Tab Controls 12 5.00 5.40
Zoom+ 7 5.00 5.00
Get the real XP look 3 5.00 5.00
Drag and Drop Tab Control 2 5.00 5.00
A high performance TCP/IP 1 5.00 5.00
BUG: MFC 7.0's CByteArray 1 5.00 5.00
Drawing Barcodes in Windows 1 5.00 5.00
SIngleton Pattern 1 5.00 5.00
Undocumented Visual C++ 121 4.96 10.33
Achieving PostScript and Wmf 25 4.96 6.93It looks that we are finally getting a new resonable formula. Best regards, Alexandru Savescu
-
If you compare the "Top Viewed Articles" with the "Top Ranked Articles", you find that the most popular articles have a relatively low rating. This makes me think that the ranking is not a good parameter, especially for the new competitions ;) I suggest this formula: Energy = rating * log10(max(10,#votes)); In this way an article with "4/5" and 50 votes (energy=6.78) is better that an article with "5/5" and 7 votes only (energy=5) The energy for the top viewed articles will be:
#votes rank energyMFC Grid control 333 4.84 12.21
The Code Project Search Bar 63 4.05 7.29
A set of ADO Classes 82 4.80 9.19
Developing a Truly Scalable 87 4.25 8.24
CButtonST v3.5 137 4.64 9.91
Visual Studio .NET Menu 26 4.38 6.19
CXImage 74 4.68 8.75
CResizableDialog 53 4.53 7.81
Cool Owner Drawn Menus 95 4.82 9.53
Layout Manager for Dialogs 94 4.85 9.57The energy for the top ranked articles will be:
#votes rank energyCustom Tab Controls 12 5.00 5.40
Zoom+ 7 5.00 5.00
Get the real XP look 3 5.00 5.00
Drag and Drop Tab Control 2 5.00 5.00
A high performance TCP/IP 1 5.00 5.00
BUG: MFC 7.0's CByteArray 1 5.00 5.00
Drawing Barcodes in Windows 1 5.00 5.00
SIngleton Pattern 1 5.00 5.00
Undocumented Visual C++ 121 4.96 10.33
Achieving PostScript and Wmf 25 4.96 6.93Hello Davide, I see that you are doing some R n D on article ratings. I have a humble request. Can you do a case study of my articles. I have always complained to Chris about low ratings and I honestly feel some of it is artificially fabricated! I have 50+ articles and that should give you a wide enough range for comparisons. Most of my articles have good view counts but very poor ratings. Some of them are in the 2s and 3s and I have wondered why warm regards Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
-
Hello Davide, I see that you are doing some R n D on article ratings. I have a humble request. Can you do a case study of my articles. I have always complained to Chris about low ratings and I honestly feel some of it is artificially fabricated! I have 50+ articles and that should give you a wide enough range for comparisons. Most of my articles have good view counts but very poor ratings. Some of them are in the 2s and 3s and I have wondered why warm regards Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
A case study on your articles based purely on page views, ratings and vote counts won't tell you anything. There are dozens of articles with high vote and view counts with deservedly low ratings, while just as many articles with viewership, high votes and undeserved low ratings. The rating system has been changed to stop multiple votes and to weight votes by membership level. Things will sort themselves out eventually. cheers, Chris Maunder he was a VB programmer, but he got better - Christian Graus
-
A case study on your articles based purely on page views, ratings and vote counts won't tell you anything. There are dozens of articles with high vote and view counts with deservedly low ratings, while just as many articles with viewership, high votes and undeserved low ratings. The rating system has been changed to stop multiple votes and to weight votes by membership level. Things will sort themselves out eventually. cheers, Chris Maunder he was a VB programmer, but he got better - Christian Graus
Chris Maunder wrote: Things will sort themselves out eventually. I hope so too. Anyway I've noticed that all my articles posted after you modified the rating system are a lot higher rated than my older ones. So I guess, that shows two things, (1) The rating system has got better and (2) My articles are improving, the improvement might be small, but it's still there ;-) Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
-
Hello Davide, I see that you are doing some R n D on article ratings. I have a humble request. Can you do a case study of my articles. I have always complained to Chris about low ratings and I honestly feel some of it is artificially fabricated! I have 50+ articles and that should give you a wide enough range for comparisons. Most of my articles have good view counts but very poor ratings. Some of them are in the 2s and 3s and I have wondered why warm regards Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
Hi, The results for the 55 articles are:
overall rating = 3.37 (low)
max energy = 7.60 (good energy - article: Some handy dialog box tricks...)
mean energy = 4.32 (low energy)
articles for beginners = 78%
articles with downloads = 33%Not to be a teacher, or a guru, but I found some common defects in the articles: - Too much articles for beginners: the essay is easy to understand, so the skilled CPians give a low rating. - Too much articles without a playable demo: compile, run, and see what happens is easier than read a good article. - Too much preformatted monolithic blocks: interleaving some text between the source code gives a better appearance. Anyway the overall rating is really too low :confused: :~ , I have another suggestion for Chris: when an author updates his articles, he could choose to reset the rating (with a check-box in the submission wizard, or asking directly to Chris): in this way he loses all the votes (bad and good) and the article has a new chance to climb the charts.
-
Hi, The results for the 55 articles are:
overall rating = 3.37 (low)
max energy = 7.60 (good energy - article: Some handy dialog box tricks...)
mean energy = 4.32 (low energy)
articles for beginners = 78%
articles with downloads = 33%Not to be a teacher, or a guru, but I found some common defects in the articles: - Too much articles for beginners: the essay is easy to understand, so the skilled CPians give a low rating. - Too much articles without a playable demo: compile, run, and see what happens is easier than read a good article. - Too much preformatted monolithic blocks: interleaving some text between the source code gives a better appearance. Anyway the overall rating is really too low :confused: :~ , I have another suggestion for Chris: when an author updates his articles, he could choose to reset the rating (with a check-box in the submission wizard, or asking directly to Chris): in this way he loses all the votes (bad and good) and the article has a new chance to climb the charts.
Hello Davide, I am thankful that you have done a case study for me :rose: Davide Pizzolato wrote: - Too much articles for beginners: the essay is easy to understand, so the skilled CPians give a low rating. Yeah, I have realized that by now that beginner level articles are not taken too well down here. Not just from the ratings, but also from the comments! Unless it is on a recent topic so that the majority of people reading it actually gain some information. Davide Pizzolato wrote: Too much articles without a playable demo: compile, run, and see what happens is easier than read a good article. Some of them cannot have a demo of any sort, but I guess you are right. People prefer a downloadable project file zipped rather than code they can copy/paste/modify to compile. Thanks for this suggestion. I had not thought of that before. Davide Pizzolato wrote: - Too much preformatted monolithic blocks: interleaving some text between the source code gives a better appearance. You mean that my sample source code is not spaced properly, eh? Thanks for that too! Anyhow if I actually give them a zipped project file I won't even have to put a lot of code in the actual article, unless they are short samples. Thanks for your valuable tips and pointers Davide. Davide Pizzolato wrote: Anyway the overall rating is really too low Yeah, puzzling as it may sound, that's true. Even allowing for an unpopular collection of articles the rating is till too low as I had voiced previously. But Chris M's new rating system has actually improved things quite a good bit. Regards, Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
-
Hello Davide, I am thankful that you have done a case study for me :rose: Davide Pizzolato wrote: - Too much articles for beginners: the essay is easy to understand, so the skilled CPians give a low rating. Yeah, I have realized that by now that beginner level articles are not taken too well down here. Not just from the ratings, but also from the comments! Unless it is on a recent topic so that the majority of people reading it actually gain some information. Davide Pizzolato wrote: Too much articles without a playable demo: compile, run, and see what happens is easier than read a good article. Some of them cannot have a demo of any sort, but I guess you are right. People prefer a downloadable project file zipped rather than code they can copy/paste/modify to compile. Thanks for this suggestion. I had not thought of that before. Davide Pizzolato wrote: - Too much preformatted monolithic blocks: interleaving some text between the source code gives a better appearance. You mean that my sample source code is not spaced properly, eh? Thanks for that too! Anyhow if I actually give them a zipped project file I won't even have to put a lot of code in the actual article, unless they are short samples. Thanks for your valuable tips and pointers Davide. Davide Pizzolato wrote: Anyway the overall rating is really too low Yeah, puzzling as it may sound, that's true. Even allowing for an unpopular collection of articles the rating is till too low as I had voiced previously. But Chris M's new rating system has actually improved things quite a good bit. Regards, Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
Nish,I wanna say happy birthday to you man.Hope we see you again as an active member in CP. :) Mazy **"If I go crazy then will you still Call me Superman If I’m alive and well, will you be There holding my hand I’ll keep you by my side with My superhuman might Kryptonite"**Kryptonite-3 Doors Down
-
Hi, The results for the 55 articles are:
overall rating = 3.37 (low)
max energy = 7.60 (good energy - article: Some handy dialog box tricks...)
mean energy = 4.32 (low energy)
articles for beginners = 78%
articles with downloads = 33%Not to be a teacher, or a guru, but I found some common defects in the articles: - Too much articles for beginners: the essay is easy to understand, so the skilled CPians give a low rating. - Too much articles without a playable demo: compile, run, and see what happens is easier than read a good article. - Too much preformatted monolithic blocks: interleaving some text between the source code gives a better appearance. Anyway the overall rating is really too low :confused: :~ , I have another suggestion for Chris: when an author updates his articles, he could choose to reset the rating (with a check-box in the submission wizard, or asking directly to Chris): in this way he loses all the votes (bad and good) and the article has a new chance to climb the charts.
Davide Pizzolato wrote: suggestion for Chris: when an author updates his articles, he could choose to reset the rating (with a check-box in the submission wizard, or asking directly to Chris): in this way he loses all the votes (bad and good) and the article has a new chance to climb the charts. Good suggestion, a lot of articles start off poor but authors improve them over time and they become good articles. I second this idea. Michael :-) Logic, my dear Zoe, merely enables one to be wrong with authority. - The Doctor
-
Nish,I wanna say happy birthday to you man.Hope we see you again as an active member in CP. :) Mazy **"If I go crazy then will you still Call me Superman If I’m alive and well, will you be There holding my hand I’ll keep you by my side with My superhuman might Kryptonite"**Kryptonite-3 Doors Down
Mazdak wrote: Nish,I wanna say happy birthday to you man Thank you. Mazy, you are the first fellow to wish me this birthday :-) Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
-
Mazdak wrote: Nish,I wanna say happy birthday to you man Thank you. Mazy, you are the first fellow to wish me this birthday :-) Nish
Author of the romantic comedy Summer Love and Some more Cricket [New Win]
-
I like it (at least for high ranked articles). What about low ranking articles with lots of votes? (guess I could pull out the calculator myself but call me lazy ;)) (apologies for reformatting your post - I wanted others to see your idea more clearly) cheers, Chris Maunder
I think the popularity ratings are biased. They tilts heavily toward MFC articles and toward "cool" subjects like grids. Meanwhile, other articles that are more specialized and receive less total views (and, consequently, less total votes) receive lower "popularity". Frankly, I find it somewhat childish. It takes me back to grammar school days when we voted for the most popular boy and girl in class.
-
I think the popularity ratings are biased. They tilts heavily toward MFC articles and toward "cool" subjects like grids. Meanwhile, other articles that are more specialized and receive less total views (and, consequently, less total votes) receive lower "popularity". Frankly, I find it somewhat childish. It takes me back to grammar school days when we voted for the most popular boy and girl in class.
The whole point of the ratigs and popularity is two-fold a) to allow those who have never been to CodeProject before a chance to look over the articles that most people found useful. b) to have something that will give good articles a chance to appear on the homepage longer than would normally be the case. No, it's not perfect, and I'm always up for other ways of accomplishing these two tasks. cheers, Chris Maunder Rub your belly and pat your head simultaneously. Sometimes that helps me make sense of things - Jon Sagara
-
The whole point of the ratigs and popularity is two-fold a) to allow those who have never been to CodeProject before a chance to look over the articles that most people found useful. b) to have something that will give good articles a chance to appear on the homepage longer than would normally be the case. No, it's not perfect, and I'm always up for other ways of accomplishing these two tasks. cheers, Chris Maunder Rub your belly and pat your head simultaneously. Sometimes that helps me make sense of things - Jon Sagara
I agree that those two goals are important. You've got a tough job trying to please everybody.