Are nested sets and dictionaries anti-pattern in Python? - python

I need to create a sort of similarity matrix based on user_id values. I am currently using Pandas to store the majority of my data, but I know that iteration is very anti-pattern, so I am considering creating a set/dictionary nest to store the similarities, similar to some of the proposed structures here
I would only be storing N nearest similarities, so it would amount to something like this:
{
'user_1' : {'user_2':0.5, 'user_4':0.9, 'user_3':1.0},
'user_2' : ...
}
It would be allowing me to access a neighbourhood by doing dict_name[user_id] quite easily.
Essentially the outermost dictionary key would hold a user_id which returns another dictionary of its N closest neighbours with user_id- similarity_value key-value sets.
For more context, I'm just writing a simple KNN recommender. I am doing it from scratch as I've tried using Surpriselib and sklearn but they don't have the context-aware flexibility I require.
This seems like a reasonable way to store these values to me, but is it very anti-pythonic, or should I be looking to do this using some other structures (e.g. NumPy or Pandas or something else I don't yet know about)?

As the comment says, there is nothing inherently wrong or anti-pythonic with using (one level of) nested dictionaries and writing everything from scratch.
Performance-wise you can probably beat your self-written solution if you use an existing data structure whose API works well with the transformations/operations you intend to perform on them. Numpy/Pandas only will help if your operations can be expressed as vectorized operations that operate on all (pairs of) elements along a common axis, e.g. all users in your top-level dictionary.

Related

Data Structure Recommendation, Nuclides

I'm writing a program to do various calculations involving nuclides. Some of these involve binding energies, magnetic moments, etc. Part of the program will need to be the storing of some dictionary, list, or something I'm unaware of as a novice Python programmer. I'd like to (by hand) create a set that contains Z, N, masses etc. Specifically, I'd like a structure that has multiple traits to a piece. I've thought of making a nested dictionary (maybe calling an attribute, nuclides[C14[attribute]]), but don't think this is intuitive. Here's the trickiest part, I'd like nuclides to referable by either Z and N, and by a string (e.g. nuclides['14C'] or nuclides[6,8]). As far as I know, dictionaries are only referenced by their label, so I'm not sure if dictionaries are ideal.
TL;DR
What's the best format for storing numerous of sets of integers/floats and a unique string where each set can be referenced by either it's string or pair of numbers.
An example of application, if say given 238Pu, finding the daughter nuclide and it's mass (both of which are in this table/data) from alpha decay.

More efficient use of dictionaries

I'm going to store on the order of 10,000 securities X 300 date pairs X 2 Types in some caching mechanism.
I'm assuming I'm going to use a dictionary.
Question Part 1:
Which is more efficient or Faster? Assume that I'll be generally looking up knowing a list of security IDs and the 2 dates plus type. If there is a big efficiency gain by tweaking my lookup, I'm happy to do that. Also assume I can be wasteful of memory to an extent.
Method 1: store and look up using keys that look like strings "securityID_date1_date2_type"
Method 2: store and look up using keys that look like tuples (securityID, date1, date2, type)
Method 3: store and look up using nested dictionaries of some variation mentioned in methods 1 and 2
Question Part 2:
Is there an easy and better way to do this?
It's going to depend a lot on your use case. Is lookup the only activity or will you do other things, e.g:
Iterate all keys/values? For simplicity, you wouldn't want to nest dictionaries if iteration is relatively common.
What about iterating a subset of keys with a given securityID, type, etc.? Nested dictionaries (each keyed on one or more components of your key) would be beneficial if you needed to iterate "keys" with one component having a given value.
What about if you need to iterate based on a different subset of the key components? If that's the case, plain dict is probably not the best idea; you may want relational database, either the built-in sqlite3 module or a third party module for a more "production grade" DBMS.
Aside from that, it matters quite a bit how you construct and use keys. Strings cache their hash code (and can be interned for even faster comparisons), so if you reuse a string for lookup having stored it elsewhere, it's going to be fast. But tuples are usually safer (strings constructed from multiple pieces can accidentally produce the same string from different keys if the separation between components in the string isn't well maintained). And you can easily recover the original components from a tuple, where a string would need to be parsed to recover the values. Nested dicts aren't likely to win (and require some finesse with methods like setdefault to populate properly) in a simple contest of lookup speed, so it's only when iterating a subset of the data for a single component of the key that they're likely to be beneficial.
If you want to benchmark, I'd suggest populating a dict with sample data, then use the timeit module (or ipython's %timeit magic) to test something approximating your use case. Just make sure it's a fair test, e.g. don't lookup the same key each time (using itertools.cycle to repeat a few hundred keys would work better) since dict optimizes for that scenario, and make sure the key is constructed each time, not just reused (unless reuse would be common in the real scenario) so string's caching of hash codes doesn't interfere.

Best data structures for 3D array in Python

I am developing a moving average filter for position "tracks" in Touch Designer, which implements a Python runtime. I'm new to Python and I'm unclear on the best data structures to use. The pcode is roughly:
Receive a new list of tracks formatted as:
id, posX, posY
2001, 0.54, 0.21
2002, 0.43, 0.23
...
Add incoming X and Y values to an existing data structure keyed on "id," if that id is already present
Create new entries for new ids
Remove any id entries that are not present in the incoming data
Return a moving average of X and Y values per id
Question: Would it be a good idea to do this as a hashtable where the key is id, and the values are a list of lists? eg:
ids = {2001: posVals, 2002: posVals2}
where posVals is a list of [x,y] pairs?
I think this is like a 3D array, but where I want to use id as a key for lots of operations.
Thanks
First, if the IDs are all relatively small positive integers, and the array is almost completely dense (that is, almost all IDs will exist), a dict is just adding extra overhead. On the other hand, if there are large numbers, or large gaps, or keys of different types, using a dict as a "sparse array" makes perfect sense.
Meanwhile, for the other two dimensions… how many IDs are you going to have, and how many pairs for each? If we're talking a handful of pairs per ID and a few thousands total pairs across all IDs, a list of pairs per ID is perfectly fine (although I'd probably represent each pair as a tuple rather than a list), and anything else would be adding unnecessary complexity. But if there are going to be a lot of pairs per ID, or a lot total, you may run into problems with storage, or performance.
If you can use the third-party library numpy, it can store a 2D array of numbers in much less memory than a list of pairs of numbers, and it can perform calculations like moving averages with much briefer/more readable code, and with much less CPU time. In fact, it can even store a sparse 3D array for you.
If you can only use the standard library, the array module can get you most of the same memory benefits, but without the simplicity benefits (in fact, your code becomes slightly more complex, as you have to represent the 2D array as a 1D array, old-school-C-style—although you can wrap that up pretty easily), or the time performance benefits, and it can't help with the sparseness.
Yes, this is how I would do it. It's very intuitive this way, assuming you're always looking things up by their id and don't need to sort in some other way.
Also, the terminology in Python is dict (as in dictionary), rather than hashtable.

Python hash table for fuzzy matching

I am trying to implement a data structure which allows rapid look-ups based on keys.
The python dict is great when my look-ups involve an equality
(e.g. key == somevalue translates to datadict[somevalue].
The problem is that I also need to be able to efficiently look up keys based on a more complex comparison, e.g. key > 50, or key.startswith('abc').
Obviously I can't use the same solution in both cases, but at the moment I can't figure out how to solve either case. Can anyone suggest a way of doing this?
It doesn't sound like you want a hash algorithm - instead some form of binary tree. Or even a list which you use the bisect module with. It'd be worth looking at: Python's standard library - is there a module for balanced binary tree?
Another option (depending on your data), would be to use use an in-memory sqlite3 database and create appropriate indices for possible lookups -- but you'll trade performance/memory and SQL syntax for flexibility...
Put all data items in a list.
Sort the list on the key.
Use binary search to efficiently find items where key > 50 or where key.startswith('abc').
Of course, this only pays off if you have really very many data items. If you have not so many, simply loop through the list and apply your condition to every key.

Need an algorithm to optimize grouping python dictionaries in hierarchical form

Previously I asked a question on how to group dictionaries in a list in hierarchical structure. Grouping Python dictionaries in hierachical form
This method works fine, but I now need to optimize. The dictionaries has a big memory overhead and this grouping gets pretty slow when we are dealing with huge number of records.
I was wondering if there is any algorithm I could use to reduce the time needed to do the grouping.
P.S: I wouldn't mind changing the data-structure as long as it gives me the same hierarchical format.
Thanks.

Categories