Having used the software for building recommender systems, this reviewer finds it to be an excellent open-source engine with useful pre-loaded datasets for quick learning. The documentation, however, is outdated and insufficient, hindering the user experience.
View Full ReviewCrab is a open source customized recommender engine creator which helps one to build recommender engines on the go without any high level knowledge of recommender engines.
A recommender engine can be built within few steps -
1. Load a dataset
2. Build a model using the Crab API
3. Define a similarity metric.
4. Add a recommender over it, using the Crab API
and you are done!
It has high level API and also support for well known python libraries which makes it unique and easy to use. Review collected by and hosted on G2.com.
This the only thing I prefer while building a recommender engine so I basically have no dislikes for Crab but I would like one thing to be improved, the documentation. Review collected by and hosted on G2.com.
- Crab is an open source which means it's freely available and anyone can raise an issue and request a feature or may implement it himself. This is one thing that captivates me towards it.
- Also as it supports various Python libraries like pandas and numpy makes itself unique from other frameworks as there are a lot of such implementations available that don't support these libraries. Review collected by and hosted on G2.com.
Yes, it has many advantages but I have pointed out a major drawback. It supports only Python which makes its use restricted to Python developers and all the windows are closed for developers using different languages like Java, C/C++, JavaScript and so on. I find this as a prime drawback as it becomes difficult to switch languages, makes us switch frameworks too. That increases the burden of learning a new library for a new language. Review collected by and hosted on G2.com.
- The very first benefit about this library is, it is open source.
- It supports various famous python libraries used for data manipulation and visualisation like numpy, pandas and matplotlib in the core making developers tasks easier.
- The API is easy to use and it's simple to understand the functionalities using the documentation provided by Crab team. Review collected by and hosted on G2.com.
The documentations needs some improvement as there are many flaws like some API is not at all mentioned and if mentioned has very less description. Also there aren't much example available making it difficult for the new learners. They have to depend on other sources, which I think is major drawback of this library. Review collected by and hosted on G2.com.
The API is high level and one need not to have in-detail knowledge of 'how recommender engines work?' but basic knowledge of python will do. Also, they provide sample data sets where one can have hands on experience while learning. Review collected by and hosted on G2.com.
The worst thing about Crab is the documentation part. They have worked well to create such an awesome library, but have not given enough to create a good documentation. I think they should focus on this part and everything else is great! Review collected by and hosted on G2.com.
This library has state of art recommendation engine algorithms which are well optimised and ready to use at production level. Along with this I really like the ease with which I can work with this library as it supports well known python libraries like numpy and matplotlib. Review collected by and hosted on G2.com.
The documentation needs some improvement as the API descriptions are not mentioned properly and it also lacks in examples. There are no enough examples which makes it difficult to learn and one need to spend more time in understanding than in implementing. Review collected by and hosted on G2.com.
This framework is well organized and has high level API which enables developer to code recommender engines within few lines of code without any prior knowledge of how recommender engines work.
The recommender engines built are dynamic and reusable which means same model can be used to train a dynamic amount of data or even a different set of data with few tweaks.
This framework supports libraries like numpy and pandas which makes data manipulation easier comparatively and boosts the process to some extent. Review collected by and hosted on G2.com.
There is no proper documentation provided by the developer community of the project as well as very few examples are available. The community should take a step and improve on this points. This really hurdles a learner and are sincere issues to take care of. Review collected by and hosted on G2.com.
- It easily merge with the famous python libraries like numpy, pandas, e.t.c which makes work more easier.
- One can directly implement algorithms like Collaborative Filtering on the go using the Crab API.
- It also provides dataset loading utilities. Review collected by and hosted on G2.com.
The documentation is not that well maintained, in some parts it lacks explanation and even there are no such examples which helps one learn it. Review collected by and hosted on G2.com.
This library is one of its kind. It reduces all the stress of complex coding while one builds one's own customised recommender system by providing a simple and easy to use API. Review collected by and hosted on G2.com.
It has support only for Python programming language and also the documentation is not that good. Review collected by and hosted on G2.com.
- It's easy interaction with other helper libraries of python like numpy and matplotlib.
- Documentation Review collected by and hosted on G2.com.
No enough video tutorials available, even on youtube. The developers should work on the same to help the users. Review collected by and hosted on G2.com.
Get practical answers, real workflows, and honest pros and cons from the G2 community or share your insights.
Averages based on real user reviews.
1 month
3 months
5%
