【摘要】：研究目標(biāo):基于互聯(lián)網(wǎng)大數(shù)據(jù)構(gòu)建CPI輿情指數(shù)輔助預(yù)測CPI。研究方法:提出了一種構(gòu)建CPI低頻與高頻輿情指數(shù)的統(tǒng)計方法,并通過選用2006年6月至2015年12月的數(shù)據(jù)驗(yàn)證了該方法的有效性。研究發(fā)現(xiàn):相關(guān)關(guān)鍵詞的搜索熱度指標(biāo)具有領(lǐng)先CPI的預(yù)測作用,依此建立的CPI輿情指數(shù)有助于改進(jìn)CPI預(yù)測精度。研究創(chuàng)新:揭示了基于相關(guān)關(guān)鍵詞的搜索熱度指標(biāo)與CPI的非線性關(guān)系,提出了一種基于門限回歸的CPI低頻輿情指數(shù)構(gòu)建方法;使用動態(tài)因子模型估計出了CPI高頻輿情指數(shù)。研究價值:預(yù)測CPI時可輔助利用基于大數(shù)據(jù)構(gòu)建的CPI低頻與高頻輿情指數(shù)信息。
[Abstract]:Objective: to construct CPI Public opinion Index aided Prediction CPI. based on Internet big data Research methods: a statistical method of constructing CPI low frequency and high frequency public opinion index is proposed. The validity of this method is verified by selecting the data from June 2006 to December 2015. It is found that the search heat index of relevant keywords can predict the leading CPI, and the CPI public opinion index is helpful to improve the accuracy of CPI prediction. Research innovation: the nonlinear relationship between search heat index based on relevant keywords and CPI is revealed, a method of constructing CPI low frequency public opinion index based on threshold regression is proposed, and the CPI high frequency public opinion index is estimated by using dynamic factor model. Research value: the low frequency and high frequency public opinion index information of CPI based on big data can be used to predict CPI.
【作者單位】： 中南財經(jīng)政法大學(xué)統(tǒng)計與數(shù)學(xué)學(xué)院;中國人民銀行長沙中心支行調(diào)查統(tǒng)計處;
【基金】：中南財經(jīng)政法大學(xué)一流學(xué)科建設(shè)項(xiàng)目“大數(shù)據(jù)統(tǒng)計預(yù)測與決策方法研究”的資助
【分類號】：F49;F726

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 彭軍;;淺談中國電信業(yè)與中國CPI[J];中國集體經(jīng)濟(jì);2008年13期

，

本文編號：2209864