def centroidsToClusterList(path_centroids):
centroids = pickle_tools_Luke.returnObjectFromPickle(path_centroids)
cluster_list = []
cluster_id = 0
for centroid in centroids:
cluster_dict = {
"cluster_id" : cluster_id,
"accousticness" : centroid[0],
"danceability" : centroid[1],
"energy" : centroid[2],
"instrumentalness" : centroid[3],
"valence" : centroid[4]
}
cluster = Cluster.Cluster(cluster_dict)
cluster_list.append(cluster)
cluster_id += 1
pickle_tools_Luke.writeToPickle(cluster_list,path_clusters)
# msd_songs = pickle_tools_Luke.returnObjectFromPickle(out_path)
# spotify_songs = pickle_tools_Luke.returnObjectFromPickle(os.path.join(dir_path,'msd_data/all_song_data.txt'))
# title_list = []
# for song in spotify_songs:
# title = song[0].lower().replace("'","").replace(" ","_")
# title_list.append(title)
# for song in msd_songs:
# if song.attributes['title'].lower().replace("'","").replace(" ","_") in title_list:
# print song.attributes['title']
all_songs = []
for subdir, dirs, files in os.walk(song_lists_with_recommended):
for file in files:
filepath = subdir + os.sep + file
if '.DS_Store' in filepath:
continue
songs_with_rec = pickle_tools_Luke.returnObjectFromPickle(filepath)
for song in songs_with_rec:
all_songs.append(song)
print "added %d songs to all_songs[]" % len(songs_with_rec)
print "dumping %d songs to: %s" % (len(all_songs), "all_songs_with_rec.txt")
pickle_tools_Luke.writeToPickle(
all_songs,
os.path.join(song_lists_with_recommended, "all_songs_with_rec.txt"))
# with Song_DB() as dbase:
# song_list = hdf5_access_Luke.getSongs_MSD(msd_original_data_folder)
# out_path = '/Users/lucasjakober/Documents/Semester 9/Combined Course Project/Code/playlist_recommender/training_data/pickle/songs_msd/song_list_mined_from_hdf5.txt'
# pickle_tools_Luke.writeToPickle(song_list, out_path)
# print songs
import Song
import pickle_tools_Luke
import os
import deltaVector
dir_path = os.path.dirname(os.path.realpath(__file__))
for subdir, dirs, files in os.walk(
os.path.join(dir_path, 'training_data/pickle/normalized')):
for file in files:
filepath = subdir + os.sep + file
if '.DS_Store' in filepath:
continue
song_list = pickle_tools_Luke.returnObjectFromPickle(filepath)
agg_vect = deltaVector.findAverage(song_list)
print agg_vect
outpath = os.path.join(
os.path.join(dir_path, 'training_data/pickle/agg_normalized'),
"agg_" + os.path.basename(filepath))
pickle_tools_Luke.writeToPickle(agg_vect, outpath)
# dir_path = os.path.dirname(os.path.realpath(__file__))
# for subdir, dirs, files in os.walk(os.path.join(dir_path,'training_data/pickle/normalized')):
# for file in files:
# filepath = subdir + os.sep + file
# if '.DS_Store' in filepath:
# continue
# song_list = pickle_tools_Luke.returnObjectFromPickle(filepath)
# new_song_list = []
# for song in song_list:
import pickle_tools_Luke
import os
dir_path = os.path.dirname(os.path.realpath(__file__))
distance_dictionaries = []
sorted_lists = []
for subdir, dirs, files in os.walk(
os.path.join(dir_path, 'training_data/pickle/normalized')):
for file in files:
filepath1 = subdir + os.sep + file
if '.DS_Store' in filepath1:
continue
if 'Fitness' in filepath1:
song_list1 = pickle_tools_Luke.returnObjectFromPickle(filepath1)
for file in files:
filepath2 = subdir + os.sep + file
if '.DS_Store' in filepath2:
continue
song_list2 = pickle_tools_Luke.returnObjectFromPickle(
filepath2)
agg_vect1 = deltaVector.findAverage(song_list1)
agg_vect2 = deltaVector.findAverage(song_list2)
name1 = os.path.basename(filepath1)
name2 = os.path.basename(filepath2)
distance_dict = deltaVector.findDistance(
agg_vect1, name1, agg_vect2, name2)
distance_dictionaries.append(distance_dict)
for dictionary in distance_dictionaries:
for centroid in centroids:
cluster_dict = {
"cluster_id" : cluster_id,
"accousticness" : centroid[0],
"danceability" : centroid[1],
"energy" : centroid[2],
"instrumentalness" : centroid[3],
"valence" : centroid[4]
}
cluster = Cluster.Cluster(cluster_dict)
cluster_list.append(cluster)
cluster_id += 1
pickle_tools_Luke.writeToPickle(cluster_list,path_clusters)
with Song_DB() as dbase:
not_clustered = pickle_tools_Luke.returnObjectFromPickle(path_all_songs_no_cluster_id)
print "clustering %d songs"%len(not_clustered)
cluster_IDs_by_song_list_index, centroids = k_means_clustering_dataset.cluster_data(not_clustered)
clustered = dataset_processing.combine_cluster_song_data(not_clustered, cluster_IDs_by_song_list_index)
pickle_tools_Luke.writeToPickle(clustered,path_all_songs_clustered)
pickle_tools_Luke.writeToPickle(centroids,path_centroids)
all_songs = pickle_tools_Luke.returnObjectFromPickle(path_all_songs_clustered)
dbase.createTable_Song()
dbase.insert_Songs(all_songs)
centroidsToClusterList(path_centroids)
cluster_list = pickle_tools_Luke.returnObjectFromPickle(path_clusters)
for cluster in cluster_list:
print cluster.attributes