ml-finance-python

python scripts for finance machine learning

git clone https://9o.is/git/ml-finance-python.git

eval_vecs.py

(1884B)

      1 #!/usr/bin/env python
      2 # -*- coding: utf-8 -*-
      3 __author__ = 'Stefan Jansen'
      4 
      5 from pathlib import Path
      6 import numpy as np
      7 import pandas as pd
      8 
      9 pd.set_option('display.expand_frame_repr', False)
     10 np.random.seed(42)
     11 
     12 from pathlib import Path
     13 from time import time
     14 import warnings
     15 from collections import Counter
     16 import logging
     17 
     18 import numpy as np
     19 import pandas as pd
     20 
     21 from gensim.models import Word2Vec, KeyedVectors
     22 from gensim.models.word2vec import LineSentence
     23 
     24 ANALOGIES_PATH = Path().cwd().parent / 'data' / 'analogies' / 'analogies-en.txt'
     25 
     26 
     27 def eval_analogies(w2v, max_vocab=15000):
     28     accuracy = w2v.wv.accuracy(ANALOGIES_PATH,
     29                                restrict_vocab=max_vocab,
     30                                case_insensitive=True)
     31     return (pd.DataFrame([[c['section'],
     32                            len(c['correct']),
     33                            len(c['incorrect'])] for c in accuracy],
     34                          columns=['category', 'correct', 'incorrect'])
     35             .assign(average=lambda x:
     36     x.correct.div(x.correct.add(x.incorrect)))).fillna(0)
     37 
     38 
     39 model_path = Path('models', 'trial_5')
     40 accuracies = pd.DataFrame()
     41 totals = {}
     42 for model_file in model_path.glob('*.bin'):
     43     _, size = model_file.stem.split('_')
     44     model = KeyedVectors.load_word2vec_format(model_file.as_posix(),
     45                                               binary=True,
     46                                               unicode_errors='ignore')
     47     accuracy = eval_analogies(model).set_index('category')
     48     total = (accuracy.loc['total',
     49                           ['correct', 'incorrect']]
     50              .sum().astype(int))
     51     totals[size] = total
     52     print(size, '\t', f"{total:,d} | {accuracy.loc['total', 'average']:.2%}")
     53 
     54     accuracies[size] = accuracy.average
     55 totals = pd.Series(totals)
     56 print(totals)
     57 totals.to_csv(model_path / 'totals.csv')
     58 accuracies.to_csv(model_path / 'accuracies.csv')