diff --git a/LDA/04f-pca.py b/LDA/04f-pca.py
new file mode 100644
index 0000000..458493c
--- /dev/null
+++ b/LDA/04f-pca.py
@@ -0,0 +1,96 @@
+
+# coding: utf-8
+
+# In[29]:
+
+# Import
+import itertools
+import shelve
+import pickle
+import numpy
+import scipy
+from scipy import sparse
+import scipy.sparse
+import scipy.io
+from mlp import *
+import mlp
+import sys
+import utils
+import dill
+from collections import Counter
+from gensim.models import LdaModel
+from sklearn.decomposition import PCA
+
+
+
+# In[3]:
+
+#30_50_50_150_0.0001
+
+# In[4]:
+
+#db=shelve.open("SPELIKE_MLP_DB.shelve",writeback=True)
+origin_corps=shelve.open("{}".format(sys.argv[2]))
+in_dir = sys.argv[1]
+if len(sys.argv) > 3 :
+    features_key = sys.argv[3]
+else :
+    features_key = "LDA"
+
+out_db=shelve.open("{}/pca_scores.shelve".format(in_dir),writeback=True)
+mlp_h = [ 250, 250  ]
+mlp_loss = "categorical_crossentropy"
+mlp_dropouts = [0.25]* len(mlp_h)
+mlp_sgd = Adam(lr=0.0001)
+mlp_epochs = 3000
+mlp_batch_size = 5
+mlp_input_activation = "relu"
+mlp_output_activation="softmax"
+
+ress = []
+print 
+
+for key in origin_corps[features_key].keys() :
+    print "#########" + key + "########"
+    dev_best =[]
+    test_best = []
+    test_max = []
+    pca = PCA(n_components=200, copy=True, whiten=True)
+    x_train_big = pca.fit_transform(origin_corps[features_key][key]["TRAIN"])
+    y_train =origin_corps["LABEL"][key]["TRAIN"]
+
+
+
+    x_dev_big = pca.transform(origin_corps[features_key][key]["DEV"])
+    y_dev = origin_corps["LABEL"][key]["DEV"]
+
+    x_test_big = pca.transform(origin_corps[features_key][key]["TEST"])
+    y_test = origin_corps["LABEL"][key]["TEST"]
+    for i in range(1,200):
+        x_train = x_train_big[:,:i]
+        x_dev =  x_dev_big[:,:i]
+        x_test = x_test_big[:,:i]
+        print "xshape",x_train.shape
+        print "xdev", x_dev.shape
+        print "xtest",x_test.shape
+        res=mlp.train_mlp(x_train,y_train,
+            x_dev,y_dev,
+            x_test ,y_test,
+            mlp_h,dropouts=mlp_dropouts,sgd=mlp_sgd,
+            epochs=mlp_epochs,
+            batch_size=mlp_batch_size,
+            save_pred=False,keep_histo=False,
+            loss="categorical_crossentropy",fit_verbose=0)
+        arg_best = numpy.argmax(res[1])
+        dev_best.append(res[1][arg_best])
+        test_best.append(res[2][arg_best])
+        test_max.append(numpy.max(res[2]))
+        print dev_best[-1],test_best[-1]
+    out_db[key]=(res,(dev_best,test_best,test_max))
+    ress.append((key,dev_best,test_best,test_max))
+    out_db.sync()
+
+for el in ress :
+    print el
+out_db.close()
+origin_corps.close()