better plotting with defaults

Just have one plot function that displays: empty plot if no data, one or the other with data, and both if both have data. simpler, cleaner, smaller.

flask/plot.py

@@ -7,210 +7,53 @@
 import time, os
 
 CURRENT_YEAR=2021
+MIN_YEAR = 2000
 NEUTRAL = 0 #the line deciding which side of the issue
 IMG_PATH='./static/images/'
 
-#create plot of only SCOTUS opinion polarity given {YEAR:POLARITY}
-def scotus_plot(sc_polarity,title="ISSUE"):
-
- #SCOTUS years, liberal v. conservative sentiment
- x_sc=list(sc_polarity.keys())
- y_sc=list(sc_polarity.values())
-
- #get basic stats
- slope, intercept, r_value, p_value, std_err = stats.linregress(x_sc,y_sc)
-
- #degree 1 model for SC
- sc_deg1=np.polyfit(x_sc,y_sc,1)
- g1 = np.poly1d(sc_deg1)
- #degree 3 model for SC
- sc_deg3=np.polyfit(x_sc,y_sc,3)
- g3 = np.poly1d(sc_deg3)
-
- #find paradigm shift for SC.
- #assuming one point of intersection.
- #this is guaranteed for a linear fit.
- p_shift_sc=opt.fsolve(g1-NEUTRAL,2000)
-
- #create figure, axes
- fig = plt.figure()
- ax = fig.add_subplot(111)
-
- #supreme court decisions
- sc_scatter=ax.scatter(x_sc,y_sc,c='r',marker="+",label='Supreme Court',vmin=-1, vmax=1)
- #model for SC data
- ax.plot(x_sc,g1(x_sc),'-r')
-
- #include horizontal neutral opinion line
- #purple='#551A8B'
- ax.axhline(y=NEUTRAL,color='k')
- #include vertical lines at crossover moments
-
- ax.plot([p_shift_sc], [NEUTRAL], marker='|', markersize=7, color="black")
-
- ax.legend()
- plt.xlabel('Time (year)')
- plt.ylabel('Polarity')
- plt.title('SCOTUS Opinion Polarity on ' + title)
- plt.grid(True)
- plt.yticks(np.arange(-1,2,1),["Conservative","Neutral","Liberal"]) #y-ticks are -1, 0, +1
-
- #fig.colorbar(sc_scatter)
-
- #min_yr=min(x_sc)
- min_yr=min(x_sc)
- plt.axis([min_yr-2, CURRENT_YEAR, -1.1, 1.1])
-
- #remove previous plot and save new one with title corresponding to keywords
- plot_filename = "sc_v_po_" + title + ".png"
- for filename in os.listdir('./static/images/'):
- if filename.startswith('sc_v_po_'):
- os.remove('./static/images/' + filename)
- plt.savefig(IMG_PATH+plot_filename, bbox_inches='tight')
- #plt.show()
- plt.gcf().clear()
-
- return plot_filename
-
-#create plot of public opinion polarity given {YEAR:POLARITY}
-def public_plot(po_polarity,title="ISSUE"):
-
- #PUBLIC OPINION
- x_po=list(po_polarity.keys())
- y_po=list(po_polarity.values())
-
- #get basic stats
- slope, intercept, r_value, p_value, std_err = stats.linregress(x_po,y_po)
-
- #degree 1 model for PO
- po_deg1=np.polyfit(x_po,y_po,1)
- f1 = np.poly1d(po_deg1)
- #degree 3 model for PO
- po_deg3=np.polyfit(x_po,y_po,3)
- f3 = np.poly1d(po_deg3)
-
- #find paradigm shift for PO.
- #assuming one point of intersectionself.
- #this is guaranteed for a linear fit.
- p_shift_po=opt.fsolve(f1-NEUTRAL,2000)
-
- #create figure, axes
- fig = plt.figure()
- ax = fig.add_subplot(111)
-
-
- #average public opinion on issue
- ax.scatter(x_po,y_po,c='b',marker="x",label='Public Opinion',vmin=-1, vmax=1)
- #model for PO data
- ax.plot(x_po,f1(x_po),'-b')
-
- #include horizontal neutral opinion line
- #purple='#551A8B'
- ax.axhline(y=NEUTRAL,color='k')
- #include vertical lines at paradigm shift moments
- ax.plot([p_shift_po], [NEUTRAL], marker='|', markersize=7, color="black")
-
- ax.legend()
- plt.xlabel('Time (year)')
- plt.ylabel('Polarity')
- plt.title('[Public Opinion TBD] on ' + title)
- plt.grid(True)
- plt.yticks(np.arange(-1,2,1),["Conservative","Neutral","Liberal"]) #y-ticks are -1, 0, +1
-
- #fig.colorbar(sc_scatter)
-
- #min_yr=min(x_po)
- min_yr=min(x_po)
- plt.axis([min_yr-2, CURRENT_YEAR, -1.1, 1.1])
-
- #remove previous plot and save new one with title corresponding to keywords
- plot_filename = "sc_v_po_" + title + ".png"
- for filename in os.listdir('./static/images/'):
- if filename.startswith('sc_v_po_'):
- os.remove('./static/images/' + filename)
- plt.savefig(IMG_PATH+plot_filename, bbox_inches='tight')
- #plt.show()
- plt.gcf().clear()
-
- return plot_filename
-
 #create plot of both SCOTUS & public opinion polarity given {YEAR:POLARITY}
-def scotus_v_public_plot(sc_polarity,po_polarity,title="ISSUE"):
-
- #PUBLIC OPINION
- x_po=list(po_polarity.keys())
- y_po=list(po_polarity.values())
-
- #get basic stats
- slope, intercept, r_value, p_value, std_err = stats.linregress(x_po,y_po)
-
- #degree 1 model for PO
- po_deg1=np.polyfit(x_po,y_po,1)
- f1 = np.poly1d(po_deg1)
- #degree 3 model for PO
- po_deg3=np.polyfit(x_po,y_po,3)
- f3 = np.poly1d(po_deg3)
-
- #find paradigm shift for PO.
- #assuming one point of intersectionself.
- #this is guaranteed for a linear fit.
- p_shift_po=opt.fsolve(f1-NEUTRAL,2000)
-
- #SCOTUS years, liberal v. conservative sentiment
- x_sc=list(sc_polarity.keys())
- y_sc=list(sc_polarity.values())
-
- #get basic stats
- slope, intercept, r_value, p_value, std_err = stats.linregress(x_sc,y_sc)
-
- #degree 1 model for SC
- sc_deg1=np.polyfit(x_sc,y_sc,1)
- g1 = np.poly1d(sc_deg1)
- #degree 3 model for SC
- sc_deg3=np.polyfit(x_sc,y_sc,3)
- g3 = np.poly1d(sc_deg3)
-
- #find paradigm shift for SC.
- #assuming one point of intersection.
- #this is guaranteed for a linear fit.
- p_shift_sc=opt.fsolve(g1-NEUTRAL,2000)
+def scotus_v_public_plot(sc_polarity=None,po_polarity=None,title="ISSUE"):
 
  #create figure, axes
  fig = plt.figure()
  ax = fig.add_subplot(111)
 
+ min_year = MIN_YEAR #minimum year for t-axis on plot
 
- #average public opinion on issue
- ax.scatter(x_po,y_po,c='b',marker="x",label='Public Opinion',vmin=-1, vmax=1)
- #model for PO data
- ax.plot(x_po,f1(x_po),'-b')
-
- #supreme court decisions
- sc_scatter=ax.scatter(x_sc,y_sc,c='r',marker="+",label='Supreme Court',vmin=-1, vmax=1)
- #model for SC data
- ax.plot(x_sc,g1(x_sc),'-r')
-
- #include horizontal neutral opinion line
- #purple='#551A8B'
- ax.axhline(y=NEUTRAL,color='k')
- #include vertical lines at paradigm shift moments
-
- ax.plot([p_shift_po], [NEUTRAL], marker='|', markersize=7, color="black")
-
- ax.plot([p_shift_sc], [NEUTRAL], marker='|', markersize=7, color="black")
+ #PUBLIC OPINION
+ if po_polarity is not None:
+ x_po=list(po_polarity.keys()) #get public opinion years
+ y_po=list(po_polarity.values()) #get public opinion polarity values
+ min_year = min(x_po+[min_year]) #update minimum year from public opinion data
+ slope, intercept, r_value, p_value, std_err = stats.linregress(x_po,y_po) #get basic stats
+ po_deg1=np.polyfit(x_po,y_po,1); f1 = np.poly1d(po_deg1) #degree 1 model for PO
+ p_shift_po=opt.fsolve(f1-NEUTRAL,2000) #find "paradigm shift" for PO. assuming one point of intersectionself. this is guaranteed for a linear fit.
+ ax.plot([p_shift_po], [NEUTRAL], marker='|', markersize=7, color="black") #include vertical lines at crossover moments
+ ax.scatter(x_po,y_po,c='b',marker="x",label='Public Opinion',vmin=-1, vmax=1) #average public opinion on issue
+ ax.plot(x_po,f1(x_po),'-b') #plot model for PO data
+
+ #SCOTUS OPINIONS
+ if sc_polarity is not None:
+ x_sc=list(sc_polarity.keys()) #get scotus years
+ y_sc=list(sc_polarity.values()) #get scotus polarity values
+ min_year = min(x_sc+[min_year]) #update minimum year from scotus data
+ slope, intercept, r_value, p_value, std_err = stats.linregress(x_sc,y_sc) #get basic stats
+ sc_deg1=np.polyfit(x_sc,y_sc,1); g1 = np.poly1d(sc_deg1) #degree 1 model for SC
+ p_shift_sc=opt.fsolve(g1-NEUTRAL,2000) #find "paradigm shift" for SC assuming one point of intersection. this is guaranteed for a linear fit.
+ sc_scatter=ax.scatter(x_sc,y_sc,c='r',marker="+",label='Supreme Court',vmin=-1, vmax=1) #supreme court decisions
+ ax.plot(x_sc,g1(x_sc),'-r') #plot model for SC data
+ ax.plot([p_shift_sc], [NEUTRAL], marker='|', markersize=7, color="black") #include vertical lines at crossover moments
+
+ ax.axhline(y=NEUTRAL,color='k') #include horizontal neutral opinion line. purple='#551A8B'
 
  ax.legend()
  plt.xlabel('Time (year)')
  plt.ylabel('Polarity')
- plt.title('SCOTUS v. [Public Opinion TBD] on ' + title)
+ plt.title('SCOTUS v. Public Opinion on ' + title)
  plt.grid(True)
  plt.yticks(np.arange(-1,2,1),["Conservative","Neutral","Liberal"]) #y-ticks are -1, 0, +1
 
- #fig.colorbar(sc_scatter)
-
- #min_yr=min(x_po+x_sc)
- min_yr=min(x_sc+x_po)
- plt.axis([min_yr-2, CURRENT_YEAR, -1.1, 1.1])
+ plt.axis([min_year-2, CURRENT_YEAR, -1.1, 1.1])
 
  #remove previous plot and save new one with title corresponding to keywords
  plot_filename = "sc_v_po_" + title + ".png"

flask/scvpo.py

@@ -6,7 +6,7 @@
 from search_public_opinions_by_keyword import relevant_questions_anes_df
 from polarity_scotus import sc_polarity
 from polarity_public import po_polarity
-from plot import scotus_plot, public_plot, scotus_v_public_plot
+from plot import scotus_v_public_plot
 
 #set up Flask app
 app = Flask(__name__)
@@ -37,38 +37,29 @@ def scotus_v_public():
  keywords = [word1,word2,word3]
  #search scotus cases via tf-idf of scotus opinion corpus. returns SCDB sub-df.
  scotus_results = relevant_cases_scdb_df(keywords,scotus_vocab,scotus_tfidf_matrix_sparse,scotus_opin_id_df,all_scdb_case_data)
+ scotus_polarity = None
  #search public opinions via tf-idf of all public opinion questions. returns ANES codebook sub-df
  public_results = relevant_questions_anes_df(keywords,public_vocab,public_tfidf_matrix_sparse,public_opin_id_df,anes_codebook_df)
- #no results for either, flash "no output" and don't plot anything
- if public_results.empty and scotus_results.empty:
- flash("No relevant cases!",'scotus_output')
- flash("No relevant questions!",'public_output')
- #if only results for public opinion, flash questions and plot public opinion polarity
- if (not public_results.empty) and (scotus_results.empty):
- flash("No relevant cases!",'scotus_output')
+ public_polarity = None
+ #if results for public opinion, flash questions and compute public opinion polarity
+ if not public_results.empty:
  for case_name in public_results['question']:
  flash(case_name,'public_output')
  rel_vcf_codes = [SURVEY_YEAR_VCF_CODE]+list(public_results['vcf_code']) #get relevant vcf codes including code for survey year
  rel_ans_df = anes_response_df.filter(items=rel_vcf_codes) #filter the relevant repsonses/answers by VCF code
- plot_filename=public_plot(po_polarity=po_polarity(public_results,rel_ans_df),title="+".join(keywords))
- flash(plot_filename,'plot_filename')
+ public_polarity = po_polarity(public_results,rel_ans_df) #compute polarity of public opinion
+ else:
+ flash("No relevant questions!",'public_output')
  #if only results for scotus cases, flash cases and plot scotus polarity
- if (public_results.empty) and (not scotus_results.empty):
- flash("No relevant questions!",'public_output') #no results for PO
+ if not scotus_results.empty:
  for case_name in scotus_results['caseName']:
  flash(case_name,'scotus_output')
- plot_filename=scotus_plot(sc_polarity=sc_polarity(scotus_results),title="+".join(keywords)) #plot SCOTUS polarity only
- flash(plot_filename,'plot_filename')
- #if results for both, display both questions and cases and plot polarity
- if (not public_results.empty) and (not scotus_results.empty):
- for case_name in public_results['question']:
- flash(case_name,'public_output')
- for case_name in scotus_results['caseName']:
- flash(case_name,'scotus_output')
- rel_vcf_codes = [SURVEY_YEAR_VCF_CODE]+list(public_results['vcf_code']) #get relevant vcf codes including code for survey year
- rel_ans_df = anes_response_df.filter(items=rel_vcf_codes) #filter the relevant repsonses/answers by VCF code
- plot_filename=scotus_v_public_plot(sc_polarity=sc_polarity(scotus_results),po_polarity=po_polarity(public_results,rel_ans_df),title="+".join(keywords))
- flash(plot_filename,'plot_filename')
+ scotus_polarity = sc_polarity(scotus_results) #compute polarity of scotus opinions
+ else:
+ flash("No relevant cases!",'scotus_output')
+ #plot with results available. no results ==> empty, neutral plot.
+ plot_filename=scotus_v_public_plot(sc_polarity=scotus_polarity,po_polarity=public_polarity,title="+".join(keywords))
+ flash(plot_filename,'plot_filename')
  return redirect('/')
  return render_template("index.html",title="SCOTUS v. Public Opinion",form=form)