adega#47: Add admission_class_ira_per_semester function on admission_analysis.py

src/script/analysis/admission_analysis.py

@@ -9,6 +9,59 @@ import numpy as np
 ANO_ATUAL = 2017
 SEMESTRE_ATUAL = 2
 
+def admission_class_ira_per_semester(df):
+
+	"""
+	Calculate the average IRA in every semester of the admission classes.
+
+	This function group the dataframe by admission classes. 
+	Then group each class by semesters. 
+	And finally group each semester by student. 
+	Calculate each student's IRA and then the average IRA for the class.
+
+	Parameters
+	----------
+	df : DataFrame
+	
+	Returns
+	-------
+	dict of {list:dict}
+
+			dict_admission={
+			(admission_class1):{semester1:ira, semester2:ira, ...},
+			(admission_class2):{semester1:ira, semester2:ira, ...},
+			...
+			}
+
+	Examples
+	--------
+	{('2005', '1'): {(2012, '1o. Semestre'): 0.485, 
+					 (2007, '1o. Semestre'): 0.6186531973412296, ...} ...}
+	"""
+
+	df = df[df['SITUACAO'].isin(Situation.SITUATION_AFFECT_IRA)]
+	df = df[ df['TOTAL_CARGA_HORARIA'] != 0]
+	admission_grouped = df.groupby(['ANO_INGRESSO_y','SEMESTRE_INGRESSO'])
+	dict_admission = {}
+	
+	for admission in admission_grouped:
+	#admission_grouped is a tuple of tuples, each tuple contains 0-tuple year/semester & 1-dataframe 
+		dict_ira_semester = {}
+		semester_grouped = admission[1].groupby(['ANO','PERIODO'])
+		
+		for semester in semester_grouped:
+			student_grouped = semester[1].groupby('ID_ALUNO')
+			ira_class = 0
+			
+			for student in student_grouped:
+				#TODO: Verify if this can be calculated without groupby
+				ira_individual =( (student[1].MEDIA_FINAL*student[1].TOTAL_CARGA_HORARIA).sum() )/(100*student[1].TOTAL_CARGA_HORARIA.sum())
+				ira_class += ira_individual
+			ira_class = ira_class / len(student_grouped)
+			dict_ira_semester.update({semester[0]:ira_class})		
+		dict_admission.update({admission[0]:dict_ira_semester})
+	return dict_admission
+
 def iras_alunos_turmas_ingressos(df):
 	iras = ira_alunos(df)
 	

src/script/build_cache.py

@@ -21,7 +21,7 @@ def build_cache(dataframe,path):
                 path = path + '/'
                 generate_degree_data(path, df)
                 generate_student_data(path+'students/',df)
-                generate_admission_data(path+'/admission/',df)
+                generate_admission_data(path+'admission/',df)
                 generate_course_data(path+'disciplina/' ,dataframe)
 
 def generate_degree_data(path, dataframe):
@@ -121,33 +121,38 @@ def generate_student_list(path):
 
 def generate_admission_data(path,df):
 
-	listagem = []
+    listagem = []
 
-	analises = [
-		("ira", media_ira_turma_ingresso(df)),
-		("desvio_padrao", desvio_padrao_turma_ingresso(df)),
-	]
+    analises = [
+    	("ira", media_ira_turma_ingresso(df)),
+    	("std", desvio_padrao_turma_ingresso(df)),
+    	("ira_per_semester", admission_class_ira_per_semester(df)),
+    ]
 
-	# cria um dicionario com as analises para cada turma
-	turmas = defaultdict(dict)
-	for a in analises:
-		for x in a[1]:
-			turmas[x][ a[0] ] = a[1][x]
+# cria um dicionario com as analises para cada turma
+    turmas = defaultdict(dict)
+    for a in analises:
+        for x in a[1]:
+            valor = a[1][x]
+            x = ( str(x[0]),  str(x[1]))
+            turmas[x][ a[0] ] = valor
 
-	listagem = []
+    listagem = []
 
-	for t in turmas:
-		resumo_turma = {
-			"ano": x[0],
-			"semestre": x[1]
-		}
+    for t in turmas:
+        resumo_turma = {
+            "ano": t[0],
+            "semestre": t[1]
+        }
+
+        for analise in turmas[t]:
+            resumo_turma[analise] = turmas[t][analise]
+
+        listagem.append(resumo_turma)
 
-		for analise in turmas[t]:
-			resumo_turma[analise] = turmas[t][analise]
 
-		listagem.append(resumo_turma)
 
-	save_json(path+"lista_turma_ingresso.json", listagem)
+    save_json(path+"lista_turma_ingresso.json", listagem)
 
 
 def generate_admission_list(path,df):
@@ -164,3 +169,5 @@ def generate_course_data(path,df):
 
         disciplinas = listagem_disciplina(df,lista_disciplinas)
         save_json(path+'disciplinas.json',disciplinas)
+
+