I would like to evaluate a mahout based recommendation engine of a fashion E-Commerce Site. They use shopping card information about item bought thogether - so boolean. I want to evaluate the engine using precision and recall.
1) How can I use these metrics to evaluate the recommendation engine? Is it just possible to use these values when altering the algorithm and to then check against yourself?
2) Or does it make sense to compare to other algorithms (also using boolean data)? If yes, is there any benchmark of precision and recall available (e.g if precision is x and recall is y, then algorithm should be discarded or accepted)?
Hoping to find help I thank you in advance guys!
Well in the Information Retrieval context the items are handled in boolean manner i.e., they’re either relevant or non relevant. Mahout’s GenericRecommenderIRStatsEvaluator utilizes data splitter to make a set from already preferred (or bought in your case, bought) items which represent relevant items. In mahout’s case the selected items are top-n most preferred items. So, since the ratings are boolean it just selects n preferred items. I don’t believe this would make evaluation itself drastically more inaccurate than with normal five star ratings since buying is pretty strong sign of preference. So:
1) If you have managed to make recommendations then you are able to evaluate the recommendations using precision and recall as metrics.
2) I have used a random recommender as an benchmark (just an implementation of an mahout recommender which selects n random items). It usually produces pretty low precision and recall so if the algorithm has lower precision and recall than random recommender it probably should be ditched. Other metric that would I look in offline evaluation phase is reach, since recommender which produces recommendations only to 80 users out of 6000 active users is pretty useless.
Also it should be noted that in academic papers the precision and recall metrics have been criticized when used as a sole metric. In the end the user decides what is relevant and what is not relevant. And a recommender, which produces slightly lower than the other, is not necessarily worse than the other. For example more novel or serendipitous recommendations may lover precision and recall.
Related
At the moment, my problem has four metrics. Each of these measures something entirely different (each has different units, a different range, etc.) and each is weighted externally. I am using Drools for scoring.
I only have only one score level (SimpleLongScore) and I have to find a way to appropriately combine the individual scores of these metrics onto one long value
The most significant problem at the moment is that the range of values for the metrics can be wildly different.
So if, for example, after a move the score of a metric with a small possible range improves by, say, 10%, that could be completely dwarfed by an alternate move which improves the metric with a larger range's score by only 1% because OptaPlanner only considers the actual score value rather than the possible range of values and how changes affect them proportionally (to my knowledge).
So, is there a way to handle this cleanly which is already part of OptaPlanner that I cannot find?
Is the only feasible solution to implement Pareto scoring? Because that seems like a hack-y nightmare.
So far I have code/math to compute the best-possible and worst-possible scores for a metric that I access from within the Drools and then I can compute where in that range a move puts us, but this also feel quite hack-y and will cause issues with incremental scoring if we want to scale non-linearly within that range.
I keep coming back to thinking I should just just bite the bullet and implement Pareto scoring.
Thanks!
Take a look at #ConstraintConfiguration and #ConstraintWeight in the docs.
Also take a look at the chapter "explaning the score", which can exactly tell you which constraint had which score impact on the best solution found.
If, however, you need pareto optimization, so you need multiple best solutions that don't dominate each other, then know that OptaPlanner doesn't support that yet, but I know of 2 cases that implemented it in OptaPlanner by hacking BestSolutionRecaller.
That being said, 99% of the cases that think of pareto optimization, are 100% happy with #ConstraintWeight instead, because users don't want multiple best solutions (except during simulations), they just want one in production.
I have been reading about Finagle and trying to understand the code to figure out how Aperture's subset choice works.
I have seen that ApertureLeastLoaded has a "useDeterministicOrdering" and an "EndpointFactory" which I guess should be the key points to make the decision of which clients to take in the subset.
While reading the "deterministic subsetting" section of Google SRE's book, I understood that the best way to pick a subset of servers from the client point of view, is to know the total number of clients, and a unique sequential identifier of the current client, that can be used as seed of the subset generator.
In Finagle I can't understand how this process is done (I'm not super familiar with Scala) and the documentation both on the website and in the code, explain just how the aperture paradigm works, but not very clear how the initial subset is chosen
I hope somebody can enlighten me
One of the unique properties of Aperture is that its window is sized dynamically based on a clients offered load. That is, clients have a built in controller which can expand or shrink their window at runtime. This property is important as it allows clients to operate more efficiently and better adapt to a changing environment, but it does make it more complex to achieve a uniform load distribution across servers.
To contrast, the subsetting algorithm, as proposed by the Google SRE book, suggests that operators choose a static subset size which allows a uniform load distribution to be calculated analytically but introduces another static configuration that needs to be revisited as a system evolves.
Deterministic Aperture is, to the best of our knowledge, a novel algorithm for achieving a uniform load distribution while maintaining the dynamic properties of the window sizing mentioned above. From a high level, clients construct a topology of their peer cluster (which gives them a sense of ordering and proximity) and then derive a unique per-client permutation of the servers from the topology such that each server is uniformly represented across the permutations.
We are still in the early stages of testing this in production at Twitter, but early results look very promising. After we gather more empirical results, we hope to publish some more detailed content on how the algorithm works and its properties.
I starting to develop offline recommendation system using ALS algorithm.
and I need to set a goal about system.
so I wanna know what criteria used to evaluate recommendation system.
I already know MAP (mean average precision) and improvement to baselineRmse and I would like to know: how is the performance on these criterions in modern recommendation systems to set my goal.
Back in the early days of recommenders people thought predicting ratings was a good idea. This has since proven to be nearly useless of itself. If you have enough space in a UI to show a few recommendations are you going to pick the one you think the user will pick with the highest ratings? That will always result in bad performance. Rating prediction is what RMSE was designed to measure.
MAP#k on the other hand is meant to find the predictiveness in a recommender. It measures how well the training data predicts what is in the test data. It also accounts for the ordering of recommendations. Ranking/ordering of recommendations has more recently been discovered to have a much greater effect on the effectiveness of recommendations because if you can only show a limited number they had better be the most likely to cause a user to take action.
MAP#k also takes account of ranking in the sense that if you measure MAP#1 and MAP#10, you will see decreasing MAP scores if your first recommendation was more likely to be in the test data than the 10th. This means you are ordering recommendations roughly correct.
For these reason we use MAP#k. Split the "gold standard" dataset you will use in later rests and keep the split static—something like 80%-20% will work split by random choice or by time, the most recent 20% used as the test split. Build you model on the 80%, then for each interaction in the 20% get recommendations and see if the recommendations contain the item actually interacted with in the test set. The aggregate of all these will go into the MAP#k calculation, k is based on how many recommendation you ask for.
See these references and some tools we have to do this:
Kaggle blog references python code they and we ActionML use. https://www.kaggle.com/wiki/MeanAveragePrecision
ActionML analysis python code to split data sets and run MAP#k, where we use the Kaggle function. https://github.com/actionml/analysis-tools
Just starting to play around with Neural Networks for fun after playing with some basic linear regression. I am an English teacher so don't have a math background and trying to read a book on this stuff is way over my head. I thought this would be a better avenue to get some basic questions answered (even though I suspect there is no easy answer). Just looking for some general guidance put in layman's terms. I am using a trial version of an Excel Add-In called NEURO XL. I apologize if these questions are too "elementary."
My first project is related to predicting a student's Verbal score on the SAT based on a number of test scores, GPA, practice exam scores, etc. as well as some qualitative data (gender: M=1, F=0; took SAT prep class: Y=1, N=0; plays varsity sports: Y=1, N=0).
In total, I have 21 variables that I would like to feed into the network, with the output being the actual score (200-800).
I have 9000 records of data spanning many years/students. Here are my questions:
How many records of the 9000 should I use to train the network?
1a. Should I completely randomize the selection of this training data or be more involved and make sure I include a variety of output scores and a wide range of each of the input variables?
If I split the data into an even number, say 9x1000 (or however many) and created a network for each one, then tested the results of each of these 9 on the other 8 sets to see which had the lowest MSE across the samples, would this be a valid way to "choose" the best network if I wanted to predict the scores for my incoming students (not included in this data at all)?
Since the scores on the tests that I am using as inputs vary in scale (some are on 1-100, and others 1-20 for example), should I normalize all of the inputs to their respective z-scores? When is this recommended vs not recommended?
I am predicting the actual score, but in reality, I'm NOT that concerned about the exact score but more of a range. Would my network be more accurate if I grouped the output scores into buckets and then tried to predict this number instead of the actual score?
E.g.
750-800 = 10
700-740 = 9
etc.
Is there any benefit to doing this or should I just go ahead and try to predict the exact score?
What if ALL I cared about was whether or not the score was above or below 600. Would I then just make the output 0(below 600) or 1(above 600)?
5a. I read somewhere that it's not good to use 0 and 1, but instead 0.1 and 0.9 - why is that?
5b. What about -1(below 600), 0(exactly 600), 1(above 600), would this work?
5c. Would the network always output -1, 0, 1 - or would it output fractions that I would then have to roundup or rounddown to finalize the prediction?
Once I have found the "best" network from Question #3, would I then play around with the different parameters (number of epochs, number of neurons in hidden layer, momentum, learning rate, etc.) to optimize this further?
6a. What about the Activation Function? Will Log-sigmoid do the trick or should I try the other options my software has as well (threshold, hyperbolic tangent, zero-based log-sigmoid).
6b. What is the difference between log-sigmoid and zero-based log-sigmoid?
Thanks!
First a little bit of meta content about the question itself (and not about the answers to your questions).
I have to laugh a little that you say 'I apologize if these questions are too "elementary."' and then proceed to ask the single most thorough and well thought out question I've seen as someone's first post on SO.
I wouldn't be too worried that you'll have people looking down their noses at you for asking this stuff.
This is a pretty big question in terms of the depth and range of knowledge required, especially the statistical knowledge needed and familiarity with Neural Networks.
You may want to try breaking this up into several questions distributed across the different StackExchange sites.
Off the top of my head, some of it definitely belongs on the statistics StackExchange, Cross Validated: https://stats.stackexchange.com/
You might also want to try out https://datascience.stackexchange.com/ , a beta site specifically targeting machine learning and related areas.
That said, there is some of this that I think I can help to answer.
Anything I haven't answered is something I don't feel qualified to help you with.
Question 1
How many records of the 9000 should I use to train the network? 1a. Should I completely randomize the selection of this training data or be more involved and make sure I include a variety of output scores and a wide range of each of the input variables?
Randomizing the selection of training data is probably not a good idea.
Keep in mind that truly random data includes clusters.
A random selection of students could happen to consist solely of those who scored above a 30 on the ACT exams, which could potentially result in a bias in your result.
Likewise, if you only select students whose SAT scores were below 700, the classifier you build won't have any capacity to distinguish between a student expected to score 720 and a student expected to score 780 -- they'll look the same to the classifier because it was trained without the relevant information.
You want to ensure a representative sample of your different inputs and your different outputs.
Because you're dealing with input variables that may be correlated, you shouldn't try to do anything too complex in selecting this data, or you could mistakenly introduce another bias in your inputs.
Namely, you don't want to select a training data set that consists largely of outliers.
I would recommend trying to ensure that your inputs cover all possible values for all of the variables you are observing, and all possible results for the output (the SAT scores), without constraining how these requirements are satisfied.
I'm sure there are algorithms out there designed to do exactly this, but I don't know them myself -- possibly a good question in and of itself for Cross Validated.
Question 3
Since the scores on the tests that I am using as inputs vary in scale (some are on 1-100, and others 1-20 for example), should I normalize all of the inputs to their respective z-scores? When is this recommended vs not recommended?
My understanding is that this is not recommended as the input to a Nerual Network, but I may be wrong.
The convergence of the network should handle this for you.
Every node in the network will assign a weight to its inputs, multiply them by their weights, and sum those products as a core part of its computation.
That means that every node in the network is searching for some coefficients for each of their inputs.
To do this, all inputs will be converted to numeric values -- so conditions like gender will be translated into "0=MALE,1=FEMALE" or something similar.
For example, a node's metric might look like this at a given point in time:
2*ACT_SCORE + 0*GENDER + (-5)*VARISTY_SPORTS ...
The coefficients for each values are exactly what the network is searching for as it converges.
If you change the scale of a value, like ACT_SCORE, you just change the scale of the coefficient that will be found by the reciporical of that scaling factor.
The result should still be the same.
There are other concerns in terms of accuracy (computers have limited capacity to represent small fractions) and speed that may enter this, but not being familiar with NEURO XL, I can't say whether or not they apply for this technology.
Question 4
I am predicting the actual score, but in reality, I'm NOT that concerned about the exact score but more of a range. Would my network be more accurate if I grouped the output scores into buckets and then tried to predict this number instead of the actual score?
This will reduce accuracy, although you should converge to a solution much faster with fewer possible outputs (scores).
Neural Networks actually describe very high-dimensional functions in their input variables.
If you reduce the granularity of that function's output space, you essentially state that you don't care about local minima and maxima in that function, especially around the borders between your output scores.
As a result, you are sacrificing information that may be an essential component of the "true" function that you are searching for.
I hope this has been helpful, but you really should break this question down into its many components and ask them separately on different sites -- potentially some of them do belong here on StackOverflow as well.
I'm fairly new to clustering and related topics so please forgive my questions.
I'm trying to get introduced into this area by doing some tests, and as a first experiment I'd like to create clusters on tweets based on content similarity. The basic idea for the experiment would be storing tweets on a database and periodically calculate the clustering (ie. using a cron job). Please note that the database would obtain new tweets from time to time.
Being ignorant in this field, my idea (probably naive) would be to do something like this:
1. For each new tweet in the db, extract N-grams (N=3 for example) into a set
2. Perform Jaccard similarity and compare with each of the existing clusters. If result > threshold then it would be assigned to that cluster
3. Once finished I'd get M clusters containing similar tweets
Now I see some problems with this basic approach. Let's put aside computational cost, how would the comparison between a tweet and a cluster be done? Assuming I have a tweet Tn and a cluster C1 containing T1, T4, T10 which one should I compare it to? Given that we're talking about similarity, it could well happen that sim(Tn,T1) > threshold but sim(Tn,T4) < threshold. My gut feeling tells me that something like an average should be used for the cluster, in order to avoid this problem.
Also, it could happen that sim(Tn, C1) and sim(Tn, C2) are both > threshold but similarity with C1 would be higher. In that case Tn should go to C1. This could be done brute force as well to assign the tweet to the cluster with maximum similarity.
And last of all, it's the computational issue. I've been reading a bit about minhash and it seems to be the answer to this problem, although I need to do some more research on it.
Anyway, my main question would be: could someone with experience in the area recommend me which approach should I aim to? I read some mentions about LSA and other methods, but trying to cope with everything is getting a bit overwhelming, so I'd appreciate some guiding.
From what I'm reading a tool for this would be hierarchical clustering, as it would allow regrouping of clusters whenever new data enters. Is this correct?
Please note that I'm not looking for any complicated case. My use case idea would be being able to cluster similar tweets into groups without any previous information. For example, tweets from Foursquare ("I'm checking in ..." which are similar to each other would be one case, or "My klout score is ..."). Also note that I'd like this to be language independent, so I'm not interested in having to deal with specific language issues.
It looks like to me that you are trying to address two different problems in one, i.e. "syntactic" and "semantic" clustering. They are quite different problems, expecially if you are in the realm of short-text analysis (and Twitter is the king of short-text analysis, of course).
"Syntactic" clustering means aggregating tweets that come, most likely, from the same source. Your example of Foursquare fits perfectly, but it is also common for retweets, people sharing online newspaper articles or blog posts, and many other cases. For this type of problem, using a N-gram model is almost mandatory, as you said (my experience suggests that N=2 is good for tweets, since you can find significant tweets that have as low as 3-4 features). Normalization is also an important factor here, removing RT tag, mentions, hashtags might help.
"Semantic" clustering means aggregating tweets that share the same topic. This is a much more difficult problem, and it won't likely work if you try to aggregate random sample of tweets, due to the fact that they, usually, carry too little information. These techniques might work, though, if you restrict your domain to a specific subset of tweets (i.e. the one matching a keyword, or an hashtag). LSA could be useful here, while it is useless for syntactic clusters.
Based on your observation, I think what you want is syntactic clustering. Your biggest issue, though, is the fact that you need online clustering, and not static clustering. The classical clustering algorithms that would work well in the static case (like hierarchical clustering, or union find) aren't really suited for online clustering , unless you redo the clustering from scratch every time a new tweet gets added to your database. "Averaging" the clusters to add new elements isn't a great solution according to my experience, because you need to retain all the information of every cluster member to update the "average" every time new data gets in. Also, algorithms like hierarchical clustering and union find work well because they can join pre-existant clusters if a link of similarity is found between them, and they don't simply assign a new element to the "closest" cluster, which is what you suggested to do in your post.
Algorithms like MinHash (or SimHash) are indeed more suited to online clustering, because they support the idea of "querying" for similar documents. MinHash is essentially a way to obtain pairs of documents that exceed a certain threshold of similarity (in particular, MinHash can be considered an estimator of Jaccard similarity) without having to rely on a quadratic algorithm like pairwise comparison (it is, in fact, O(nlog(n)) in time). It is, though, quadratic in space, therefore a memory-only implementation of MinHash is useful for small collections only (say 10000 tweets). In your case, though, it can be useful to save "sketches" (i.e., the set of hashes you obtain by min-hashing a tweet) of your tweets in a database to form an "index", and query the new ones against that index. You can then form a similarity graph, by adding edges between vertices (tweets) that matched the similarity query. The connected components of your graph will be your clusters.
This sounds a lot like canopy pre-clustering to me.
Essentially, each cluster is represented by the first object that started the cluster.
Objects within the outer radius join the cluster. Objects that are not within the inner radius of at least one cluster start a new cluster. This way, you get an overlapping (non-disjoint!) quantization of your dataset. Since this can drastically reduce the data size, it can be used to speed up various algorithms.
However don't expect useful results from clustering tweets. Tweet data is just to much noise. Most tweets have just a few words, too little to define a good similarity. On the other hand, you have the various retweets that are near duplicates - but trivial to detect.
So what would be a good cluster of tweets? Can this n-gram similarity actually capture this?